Table of contents

It was a great pleasure to welcome all the participants from different countries to the 2021 6th International Symposium on Advances in Electrical, Electronics and Computer Engineering (ISAEECE 2021) in Nanjing, China during March 12-14, 2021. ISAEECE 2021 is to bring together innovative academics and industrial experts in the field of electrical, electronics and computer engineering to a common forum. The primary goal of the conference is to promote research and developmental activities in electrical, electronics and computer engineering and another goal is to promote scientific information interchange between researchers, developers, engineers, students, and practitioners working all around the world. The conference will be held every year to make it an ideal platform for people to share views and experiences in electrical, electronics and computer engineering and related areas.

The conference model was divided into two sessions, including oral presentations and keynote speeches. In the first part, some scholars, whose submissions were selected as the excellent papers, were given 15 minutes to perform their oral presentations one by one. Then in the second part, keynote speakers were each allocated 30-45 minutes to hold their speeches. In this conference, we were greatly honor to have invited four outstanding researchers to serve as our Conference Chairman.

During the conference, we invited three industry elites to present their insightful speeches. The first keynote speaker, Dr. Sunil Kumar Jha, from Nanjing University of Information Science and Technology, China. He presented a speech: Data Fusion Approaches in Human Body Odor Data Mining. The objective of the present research talk is to introduce audience about the data fusion and human body odor and to demonstrate research results related to search for an optimal subset of VOCs in body odor, which can produce differentiation in an individual by using the combination of analytical methods and chemometric analysis. The second keynote speaker, Prof. Chunbo Xiu, from Tiangong University, China. He shared a speech on his specialty related to Memristive Cellular Neural Network & Its Dynamic Characteristic Analysis. He demonstrated that chaotic memristive CNN system can be used in the secure communication by the chaos synchronization based on sliding mode control. Assoc. Prof. Pavel Loskot, from Zhejiang University-University of Illinois at Urbana-Champaign Institute (ZJUI), our third keynote speakers. He presented a speech: Monte Carlo Simulations Revisited. Their insightful speeches made every participant praised this conference for disseminating useful knowledge.

We were glad to share with you that we received lots of submissions from the conference and we selected a bunch of high-quality papers and compiled them into the proceedings after rigorously reviewed them. These papers feature following topics but are not limited to: Electronics Engineering, Electrical Engineering, Computer Engineering and other related topics. All the papers have been through rigorous review and process to meet the requirements of International publication standard.

We were grateful to the International/National advisory committee, keynote speakers, session chairs, organising committee members, student volunteers and administrative assistance of the management section of University, including accounts section, digital media and publication house. Also, we were thankful to all the authors for contributing large number of papers in the conference, because of which the conference became a story of success. It was the quality of their presentations and their passion to communicate with the other participants that really make this conference series a great success.

The Organizing Committee of ISAEECE 2021

List of Committee member are available in this pdf.

All papers published in this volume of Journal of Physics: Conference Series have been peer reviewed through processes administered by the Editors. Reviews were conducted by expert referees to the professional and scientific standards expected of a proceedings journal published by IOP Publishing.

• Type of peer review: Double-blind

• Conference submission management system: AI Scholar System

• Number of submissions received: 279

• Number of submissions sent for review: 268

• Number of submissions accepted: 155

• Acceptance Rate (Number of Submissions Accepted/Number of Submissions Received X 100): 55.6%

• Average number of reviews per paper: 2

• Total number of reviewers involved: 72

• Any additional info on review process:

• Step 1. Each of selected paper will be reviewed by two/three professional experts in the related subject area.

• Step 2. Review Reports received from the experts will be judged by one of the editors either Review Reports are logical or not?

• Step 3. If not logical, then editor can assign new reviewer or can also judge at his/her own.

• Step 4. If logical, then Review Reports will be sent to authors to modify the manuscript accordingly.

• Step 5. Authors will be required to revise their papers according to the points raised.

• Step 6. Revised version will then be evaluated by the editor for the incorporation of the points raised by the reviewers.

• Step 7. Then the editor will send the revised manuscript to the reviewers again for re-evaluation.

• Step 8. If the reviewers approve the revise version of the manuscript, then will be accepted for publication.

• Contact person for queries:

Xuexia Ye

publication@keoaeic.org

AEIC Academic Exchange Information Centre

Wheat freshness is an important factor affecting wheat quality. In the current research, a total of 375 wheat samples from five different years (2016, 2017, 2018, 2019 and 2020) were identified by using the terahertz time-domain spectroscopy (THz-TDS) technology. The THz spectrum information of wheat was acquired, and the THz-TDS signals of wheat were obtained after Savitzky-Golay pretreatment. Then the data of frequency-domain spectra, refractive index spectra and absorption spectra of wheat within the range of 0.2∼1.4 THz were obtained after calculating the optical parameters. According to the experiment results, there were differences in the refractive indexes and absorption coefficients of wheat samples from different years. The refractive indexes of wheat samples from 2020 were significantly lower than that of wheat samples from other years, and the absorption coefficients of all five kinds of samples increased as the frequency increased. The study results showed that THz-TDS was of great potential for the effective determination of wheat freshness and it provided a basis for determination of wheat quality, bearing great practical significance.

In this paper, a new lossless free space method for measuring the complex relative permittivity of dielectric material by reflection method is presented. When this method is used to measure the complex relative permittivity of dielectric material, there is no multi-value problem. The measurement system used in this method consists of a horn lens antenna and a vector network analyzer(VNA). In this paper, HFSS microwave simulation software is used to simulate the complex relative permittivity measurement method proposed in this paper. The simulation results show that this method can accurately measure the complex relative permittivity of dielectric material.

The composition and performance of the material can be expressed by the permittivity. The paper introduces the measurement principle of free space method and TRL calibration principle, the calculation formula of the sample permittivity is theoretically analysed, a horn lens antenna is designed to simulate the emission of plane waves, and numerical simulation software is used for simulation and optimization. In the working frequency range, the voltage standing wave ratio of the antenna is less than 1.49, and the maximum gain is 21.8dB. A free space measurement system is established in the software to measure the sample permittivity with different thicknesses, and compare the calculated value with the actual value. The results show that the root mean square error of the real part is less than 5.3%, and the root mean square error of the imaginary part is less than 1.3%. The correctness and applicability of the formula for calculating the permittivity are also proved.

This paper adds chaos optimization based on adaptive annealing particle swarm optimization algorithm, which improves the global convergence ability and makes up for the shortcomings of traditional PSO algorithm, which is easy to fall into local optimization in addition, this paper adds simulated annealing operator, improving the diversity of better population. In addition, the micro source of microgrid is modeled, and the constraints of the model are set. The fitness of the optimized algorithm is compared with the traditional algorithm, and the objective function involved in microgrid dispatching is taken as the application layer. Finally, through data analysis, the effectiveness of the optimization algorithm for microgrid optimal economic dispatching is confirmed.

Due to the rapid development of battlefield complex electromagnetic environment research in recent years, electromagnetic pulse (EMP) environment has gradually stepped onto the stage of information war. To make low altitude flying electronic equipment engine plays its normal combat performance in the EMP radiation environment and ensure the integrity of electronic equipment, it is urgent to research in this field. In this paper, according to GJB-1389A and other standards, CST simulation software is used for simulation analysis. Through simulation, field strength and cable coupling with the airborne engine of an armed helicopter under EMP irradiation are predicted. The conclusions have a certain guiding significance of the further study of the protection against the complex electromagnetic environment of the engine.

Compared with the tradition cooling technology, the evaporative cooling technology has the advantage of energy-saving, safety and reliability. So a self-circulating evaporative cooling structure was proposed. The narrow flow channels were designed between disk-type windings. The evaporative coolant flows in the narrow flow channels and takes away the heat loss through phase transition. In this paper the resistance characteristics of the system were studied through experiment based on the 900kVA transformer prototype.

After class, there are frequent cases of people walking but still lighting. In order to reduce the occurrence of this phenomenon, this paper proposes a classroom lighting energysaving system based on YOLOv3. The system uses the installed monitoring equipment in the existing public area as the information collector, controls the lamps through the wireless control switch, and uses the Internet as a bridge to intelligently control the lamps. The system can judge the area where the lights need to be turned off and automatically turn off the equipment in the case of people leaving the lights. While ensuring lighting, the system not only prolongs the service life of lamps, but also reduces the cost of school funds. When the system is popularized to the country, it will save a lot of electric energy, and the future development force cannot be underestimated. It is of great significance to carry out energy conservation and emission reduction in China.

Radio frequency interference (RFI) suppression has become an essential step in synthetic aperture radar (SAR) imaging processing. In this paper, a spectral filling filtering method for RFI suppression referred to notch filter is derived and analysed. The derivation results show the reason for the "residual terms" after suppression processing in frequency domain and the influence of the "residual terms". And a conclusion is drawn: the essence of interference suppression based on spectral filling filter is the dynamic change process of "residual terms". Since they cannot be completely eliminated in theory, they will affect the image quality after interference suppression, but the effect of spectral filling method is better than that of the notch filter method. The conclusion is verified by the simulation experiment.

The harmonic amplitude may exceed the limit when the large scale of wind farm is connected to the weak AC system through AC submarine cable. The harmonic with high amplitude will endanger the safety of power equipment, and even affect the normal integration of large-scale wind farm. This paper takes an actual offshore wind power connected to the system through AC submarine cable as an example. On the basis of building the model of the system in the PSCAD/EMTDC simulation software. The simulation calculation is systematically carried out. The effect of multiple factors on the harmonic characteristics is analyzed. The preliminary explanation for the harmonic amplitude out of limit is given for the actual large scale of wind farm connected to the weak AC system through AC submarine cable.

It is very important to predict the photovoltaic power generation because of the great challenges to the safe and stable operation of the power grid. Based on the analysis of traditional forecasting models, a power generation forecasting model based on rough set and neural network is proposed. Firstly, the Hampel filtering algorithm and rough set theory are used to process outliers and redundant data for the collected information. The above algorithms solve the problem of big data processing before forecasting. Finally, the fuzzy C-means clustering was used to divide the data set and combined with the neural network to predict. The key weather feature variables combined with the forecast time were taken as the input and the power was taken as the output variable. The results show that the prediction method proposed in this paper is more accurate and fast in predicting the power generation at each time.

In the ever-changing world of technology, various communication technologies have penetrated people's daily lives, which has greatly improved people's work efficiency and quality of life. By means of the Internet and various LAN technologies, users can remotely control their smart home appliances. In addition, the commercial use of 5G networks has contributed to the integration of communication technologies with people's work and life. Meanwhile, each communication technology has its own advantages and disadvantages. For wireless network technologies, distance and power consumption are always contradictory. LPWAN (Low-Power Wide-Area Network) is a wireless network technology designed to allow long-range communication with low power consumption. Therefore, this paper studied key LPWAN technologies with a focus on the LoRa technology.

After large-scale access to distributed generation of new energy generation in urban AC/DC distribution network, its own randomness and uncertainty affect the supply-demand balance mechanism of distribution network. Therefore, this paper studies the power exchange flexibility optimization method of urban AC/DC distribution network to provide reference for the operation and scheduling of distribution network and main network. First, the active power exchange model of distribution network is established, the extreme value of power exchange of distribution network is obtained, and the flexible range of exchange power is calculated. Next, the flexibility evaluation index is established, and the operation characteristics of energy storage are transformed to complete the research on power exchange flexibility optimization of urban AC/DC distribution network. Through the actual example analysis, the results show that after the optimization of the design method, the net load curve is smoother, the effect of reducing the peak valley difference is more significant, and the overall curve is relatively smooth, which shows that the design method has certain effectiveness.

Due to the existence of cogging torque, magnet flux harmonics and inverter dead time, permanent magnet synchronous motor (PMSM) has torque ripple during operation, which limits its application in high-precision situations. Stator current optimization is an easy-to-apply and accurate torque ripple suppression approach. However, existing methods fail to consider the effect of magenet flux phase angle on optimal current calculation. In this paper, a torque ripple minimization method considering harmonic magnet flux phase is presented, which uses torque waveform of an electric angular period as the optimization object. By deducing the electromagnetic torque model considering harmonic magnet flux, the torque is expressed as a dense sequence. Based on this, the objective function including torque ripple and losses minimization is constructed and optimized by genetic algorithm (GA) to obtain the optimal harmonic current. Experiment results show that the proposed approach can calculate the optimal harmonic current accurately and suppress the torque ripple effectively.

In order to improve the accuracy of wind power DC microgrid scheduling, especially in the case of LSTM prediction errors, this paper proposes a WOA optimized DC microgrid energy scheduling control strategy based on LSTM prediction algorithm. MATLAB software is used to simulate the scheduling process. Compared with the traditional LSTM-PID method, the WOA scheduling control strategy reduces the average absolute deviation of the SOC from 0.7043 to 0.1755, a reduction of 75.08%. While the median absolute deviation decreases from 0.5812 to 0.1316, a reduction of 77.36%.

With the widespread application of stable high-voltage power cables, cable failures have gradually become the main source of major accidents in the process of power transmission. The most easily damaged part is the cable joint. Since the cable joints are made on site, it is easy to introduce high-voltage cable defects and cause insulation failures. Therefore, the defect detection of high-voltage voltage joints is particularly important. In order to promote the process of urbanization, ensure people's life quality and personal safety, the popularization of high-voltage cables is an inevitable trend. However, difficulties will be encountered in the detection of cable joint faults. To improve the detection efficiency, relatively extensive research has been carried out on the detection of high-voltage cable joints at home and abroad. Based on the high-speed light sensing and pressure wave method, the online detection method of cable joint defects proposed in this paper has accurate, fast, convenient and non-contact online detection features.

Container technology has a series of advantages such as low physical resource consumption, fast startup speed, high concurrency, and can run in a variety of environments. It is widely used in scenarios such as big data and cloud computing. Container technology has certain advantages in performance, but there are some shortcomings in security. The container technology shares the kernel with the host, and its security mainly depends on the host. Once the attacker breaks through the host's defense, he can easily access the files deployed in the container, steal or tamper with the file data, and cause losses to users and users. In response to the above problems, this paper proposes a container-oriented isolation control technology, which realizes further isolation of files inside the container by adding domain names to programs and files. If the program domain name matches the file part, the files in the current container cannot be accessed, and the security of the files in the container can be effectively ensured after the host is compromised.

With the continuous development of science and technology, PLC technology has been applied in many industries. This also includes electrical automation control, which can better maintain the accuracy and stability of electrical automation control, and its development prospects are very broad. This article summarizes the application significance and advantages of PC technology in electrical automation control based on previous work experience. The author discusses the practical application of PLC technology in electrical automation from six aspects: air conditioning, coal mining, numerical control, transportation, power systems, and lathe electrical control systems.

In order to analyze the adhesion vibration characteristics of vertically arranged double bundled conductors, the vibration equation of adhesion conductor under the coupling action of aerodynamic force and electromagnetic force is established. Firstly, the electromagnetic force on the conductor with different splitting spacing is simulated by COMSOL software. Then, based on the Runge-Kutta numerical calculation method and the vibration mathematical model, the displacement time history curve and amplitude value of the adhesive wire under different wind speed, span and initial spacing are obtained. Finally, a 220kV laboratory model is built, and the real-time monitoring of the wire vibration data is carried out to verify the accuracy of the mathematical model and simulation results. The results show that the coupling effect of aerodynamic force and electromagnetic force is the main reason for the vibration of the wire. The larger the aerodynamic force and electromagnetic force, the larger the amplitude of the wire.

The hydro-dominant characteristic is notable after the asynchronous operation in Southwest China Power Grid which brings worse transient frequency deviation caused by water hammer effect. To prevent the possible cascading failure due to the worse transient frequency deviation in the secondary frequency regulation, the soft guide vane control law is proposed to suppress the water hammer effect in the transient regulation process based on the detailed turbine-penstock model. The discrete state space expression of the turbine-penstock model is built to simulate the regulation effect in the time domain and the PSO (Particle Swarm Optimization) algorithm is applied to optimize the soft guide vane control adjustment factor considering multi-object. Finally, the emergency AGC (Automatic Generation Control) adjustment in the single machine with load transmission system is conducted to verify the effect of transient frequency deviation suppression by optimized soft guide vane control.

In view of the randomness of component fault and load, considering the deficiency of traditional single reliability index, this paper introduces the Conditional Value-at-risk method based on Monte Carlo simulation, and constructs the probability model of risk assessment considering the forced outage rate of unit and line. In order to quantify uncertainty caused by random factors, the value-at-risk and conditional value-at-risk of power system security was proposed and used as a grid security index in the model combined with reliability index, which can accurately describe secure operation level in power system under different operation parameters, and reflect the severity of the accident and weakness and the grid, for further research on safe operation of power system to offer reference and decision-making basis.

Distributed energy has the limitation of "point-to-point" supply. The regional Energy Internet is a multi-dimensional energy network system with energy producers and consumers as nodes, electricity, gas, cold and hot energy networks as links, connecting the production, transmission, distribution, storage and use of energy. The regional Energy Internet realizes the full utilization of distributed energy through the operation mode of multi-source complementary, "energy network load storage" coordination, and optimized control technology.

In order to effectively improve the computational accuracy and robustness of state estimation, a power system state estimation method based on partial least square (PLS) and extreme learning machine (ELM) is proposed in this paper, which combines artificial intelligence technology with grid data. In order to solve the problem of strong correlation between measurements, PLS is used to extract important information and select variables for each measurement, and the optimal variables are input into the ELM model, thus the PLS-ELM model of state quantities is established. Finally, this method is compared with other methods. Experimental results show that the proposed method reduces the complexity of the model, and has high estimation accuracy and strong robustness.

After the power terminal network is attacked, it is necessary to adopt effective security strategies to block the attack in time to prevent further expansion of the attack damage. At this time, not only the availability of the system must be considered, but also the impact of the attack must be minimized. Traditional network isolation methods may cause excessive protection problems such as excessive isolation range and high isolation strength, and affect the continuity of power system business. Therefore, this paper conducts research on the attack penetration prevention techniques of power terminal network, under the condition of minimally affecting the real-time business continuity of the power grid, blocking low-risk terminals' control commands and/or isolating high-risk terminals' communication to realize accurate attack penetration prevention.

This paper studies the application of the exclusive energy controller for comprehensive energy consumption. Users' various energy consumption data such as water, gas, electricity, heat, cold are collected by controller. Various energy consumption information collection and sharing of park, community, building and other types of users can be realized by using data optimization and data fusion technology. The coupling optimization model is designed and the optimization strategy is studied, so as to realize the comprehensive state perception of energy consumption equipment and intelligent optimization control of comprehensive energy consumption.

The stator and rotor of the bearingless switched reluctance motor are both double salient pole structures, when the traditional current chopping control method is used, the torque and levitation force of the motor will have large fluctuations. To solve this problem, this paper studies a direct instantaneous torque control and direct levitation force control (DITC&DFC) method suitable for single-winding bearingless switched reluctance motors. This method takes the instantaneous torque and the instantaneous levitation force of the motor as the directly controlled object, eliminates the current loop, and simplifies the control algorithm while suppressing the fluctuation of the motor torque and levitation force. A 12/8-pole singlewinding bearingless switched reluctance motor with wide rotor teeth drive system is constructed, the Matlab/Simulink system simulation verifies the effectiveness and feasibility of the method.

In this paper, several different magnetic shielding schemes are proposed, the finite element static magnetic field analysis method is introduced, and the finite element method is used to simulate these methods. According to the calculation results, we analysed different magnetic shielding schemes, the results show that the opening will significantly reduce the magnetic shielding efficiency; under the same materials, the multi-layer magnetic shielding effect is significantly improved; when the opening is parallel to the direction of the magnetic induction intensity, the magnetic shielding has a better effect; when the direction of the magnetic field is unknown, the two-layer magnetic shielding opening direction can be orthogonal to improve the magnetic shielding effect.

There are a large number of capacitors in power electronic transformer (PET). If PET is started by conventional closed-loop control directly to establish DC capacitor voltage, the over-current that may lead to device damage may be caused. To solve this problem, a soft start strategy of PET based on energy feedback is proposed. In this strategy, the DC capacity of the low voltage side is moderately overcharged at first, and then part of the energy is fed back to the high voltage side to make the voltage of the high voltage DC side reach the reference value. Finally, the CHB is allowed to enter the closed-loop control, thus completely eliminating the impulse current of the AC side. This method avoids the complex parameter design process, and completely relies on the theoretical model rather than experience to guide the design of the process of soft start. Simulation results verify the feasibility and effectiveness of the proposed method.

Due to the existence of power converter dead time, flux harmonics and other nonlinear factors, the motor phase current contains a large number of high-order harmonics, so the electromagnetic torque generated by the interaction of current and flux also contains various orders of harmonics. In order to eliminate current harmonics, mathematical morphology filter is proposed to suppress current harmonics. However, the single-scale morphological filtering has shortcomings in preserving information details. In order to make up for the deficiency of the traditional single-scale morphological filtering in preserving information details, a weighted multi-scale mathematical morphology method is proposed to suppress the current harmonics.

Traditional proportional feed-forward strategy can well restrain harmonic disturbance from grid background, but in the weak grid, positive feedback channel and grid-connected current inner-loop are coupling with each other by grid impedance, and robustness is substantially decreased. According to the above situation, a kind of new feed-forward control strategy is proposed, and adding more resonant link on positive feedback channel of point of common coupling voltage to change full pass characteristics of positive feedback channel which only make positive feedback channel have feedback effect on the frequency of main grid background harmonics. Inverter can both have great robustness and harmonic suppression ability in weak grid. In the end, the effectiveness and correctness of proposed control strategy is verified by the simulation model of single-phase grid-connected inverter.

In this paper, two types of ellipse-pattern broadband metamaterial absorbers with high absorptivity are designed and analyzed. The single-ellipse periodic absorber can achieve more than 90% absorption in the range of 8.8 to 12.5 μm, and the average absorption in this range is 96.5%. Further exploration of its electromagnetic field's distribution revealed that the coupling of propagating surface plasmon resonance and localized surface plasmon resonance is responsible for its character. An optimized four-ellipse periodic absorber can achieve 89.5% average absorption in the range of 8 to 15 μm, and the absorption is insensitive to the incident angle. The variations of absorption of the absorber under different polarization and incident angles are also discussed. This design has potential application value in uncooled infrared detectors and other fields.

The electric reactor will generate irregular electromagnetic vibration during the onsite operation of substation. If so for a long time, it will cause problems such as broken indoor floor and loose connecting parts, endanger the building structure or damage to the building, thus leading to waste of land resources in indoor substation. In order to effectively suppress the irregular electromagnetic vibration of the electric reactor, this paper proposed an adaptive damper system of the electric reactor based on magnetorheological fluid to realize the all-around vibration suppression of reactor field vibration. This system placed the homemade damper in the center of each surface of the reactor, and used external filter circuit and signal amplification circuit to process the vibration signal and transform it into electrical signal, which controlled the single-chip microcomputer to regulate the damper to output the damping force, forming an adaptive damper system based on the magnetorheological fluid. Then, the good performance of vibration reduction of the system was verified by the onsite vibration reduction test of the reactor. And the adaptive regulate ability of the system is superior and is of good engineering application value.

Aiming at the problem that the transmission efficiency of SS-compensated MCR-WPT (Magnetic resonance wireless power transmission) is affected by load changes, this paper uses mutual inductance theory to analyze the equivalent circuit model, obtains the relationship between load and transmission efficiency, and proposes a controllable rectifier bridge for load matching, the front-end cascaded buck converter makes the output voltage regulation control scheme. Finally, modeling and simulation are used to verify the correctness of the load matching theory and control strategy.

Accurate detection of solar irradiance has important application value for theoretical research on climate change and atmospheric analysis, so this paper proposes the integrated spectral sun-photometer to enrich methods of obtaining the solar irradiance and entire atmospheric transmittance. With the ARM microprocessor as central processing unit and Linux OS as the software platform, the two-dimensional turntable and grating spectrometer are combined to construct sun-photometer which can measure continuous wavebands. This article advances the solution scheme to realize data transmission between microcontroller and turntable through CANopen communication. Meanwhile, the spectrometer based on USB communication protocol is used to obtain spectrum from visible to near-infrared band. In addition, both the multiple task operating system Linux and application framework Qt are employed to design friendly user-interface and display real-time process images, which would simplify and integrate the solar radiometer. After Langley calibration is used to calculate the constants to achieve calibration of instrument in spectral band, measured data of the radiometer and results of the DTF solar radiometer are compared and analysed, which confirms the reliability of integrated sun-photometer.

In order to enhance the stable operation of the multi-energy complementary microgrid for wind, solar, and diesel storage, reduce operating costs, and solve the problems of large randomness, low accuracy, and slow convergence of traditional microgrid optimization multi-objective decision-making, a differential evolution based on the entropy weight method to determine the weight is proposed. Firstly, we establish a microgrid integrated energy model from the perspectives of system operation stability, economy and environmental protection; combined with the Pareto optimal solution set, multi-objectives are weighted according to the entropy weight method, and the multi-objective optimization problem is transformed into single-objective optimization The problem is to avoid artificial setting of weight factors; the calculation example shows that this method is more economical and reasonable in optimization results, and provides an economic, reliable and environmentally friendly microgrid configuration strategy for users to increase power capacity.

In this paper, an electric filed focus scheme based on phase adjustable microwave source is used to improve the power transmission efficiency. A microwave power transmission system is consisted of the microwave source, a transmitting antenna and a receiving antenna. Measurement results proved that when compared with the same phases condition, the output DC power of the receiving antenna is increased from 23.5W to 31.5W and the relative increment of power is 34% under the focal phase condition. In addition, the system-level MW-DC power transmission efficiency is increased by 2.7%.

In order to study the effects of hydrogen input molar flow, oxygen input molar flow, water vapor input molar flow and real-time temperature on the output voltage of solid oxide fuel cell (SOFC), this paper proposes a method to model SOFC using a BP neural network optimized by immunogenetic algorithm (IA). MATLAB simulation results show that the SOFC model established in this paper can more accurately reflect the actual SOFC input and output characteristics. And the model has a high fit to the real data.

In this paper, the simulation analysis and research on the starting and generating system based on two-stage brushless synchronous motor and permanent magnet synchronous motor are respectively carried out, and the brushless synchronous motor AC/DC mixed excitation scheme is proposed. The starting operation adopts vector control, and the output voltage precision is improved by adjusting the excitation voltage of the exciter. A permanent magnet synchronous motor starting and generating system is constructed, and the control of starting and generating is realized by using vector control based on SVPWM, the results show that the permanent magnet synchronous motor has fast starting speed, good dynamic and static performance, high power supply quality and simple hardware composition, which reduces the complexity of the control system and is convenient for engineering implementation.

At present, there is no research on nanosecond displacement measurement technology for spacecraft thermal test in China. Firstly, the principle of laser displacement sensor measurement is introduced. A system of hardware and software design and implementation was designed. Finally, a set of test system is built to verify the application of laser displacement meter in the spacecraft vacuum thermal test displacement measurement system. The experimental results and data analysis show that the laser displacement meter can meet the nanosecond displacement measurement requirements of spacecraft in vacuum and low temperature environment.

Energy storage is the key technology for large-scale renewable energy access and energy Internet construction. Among the numerous energy storage technologies, the pumped storage and electrochemical energy storage are the fastest developing. In the paper, the present application status of the pumped storage and electrochemical energy storage is introduced. In this paper, the development prospect of energy storage and its impact on power system are analyzed, and the overall strategies and ideas to deal with the development of energy storage are put forward.

In recent years, the massive increase in electric vehicles (EV) and renewable energy (RER) has caused serious problems in the power system network. During peak hours of electricity consumption, the soaring demand for electricity from variable renewable energy sources severely undermines the stability and efficiency of the power supply. This article first reviews the infrastructure and charging strategies of electric vehicle batteries. After modeling and analyzing the charging activities, the results show that the charging behavior can transform the charging demand from the peak time period to the off-peak time period, and the online MSS optimization of electric vehicle charging is extracted through the Jacobian matrix. Finally, it provides a practical overview of the existing and future solutions to manage the large-scale integration of EVs into the network.

With the popularization and application of "Internet of Things" technologies such as equipment online monitoring and intelligent inspection in the field of substation operation and maintenance, the number of subsystems in smart substations has increased rapidly, and the types of information and the amount of information have shown explosive growth. However, due to disorder and irregular expression, the potential value of a large amount of information is difficult to be effectively used. This paper proposes a "system code-device code-information code" information coding technology that serves the operation and maintenance of information equipment to make information orderly and standardized. This technology lays the foundation for quickly and accurately finding the required information, fully mining and effectively using the value of information, strengthening information fusion between subsystems, and realizing digital management and control of substation equipment information.

Carrying out research on the aging and burning performance of cable insulation sheath is an important means to prevent cable fires. The thermal aging test box, cone calorimeter and synchronous thermal analyzer were used to study the effect of thermal aging on the combustion performance of AC 380V ZR-YJV cable insulation sheath. The results show that in the early stage of thermal aging of the cable, the ignition time of the YJV cable insulation sheath material gradually increases; compared with unaged cables, the ignition time of cables being aged for 15 days has increased from 8 seconds to 57 seconds while the initial peak value of the heat release rate of the cable decreased from 123 kW/m² to 22 kW/m². A large amount of easily decomposable additives in the early stage of thermal volatilized from aging cable sheath may be responsible for the reduction trend of the ignition and fire occurrence of the cable insulation sheath. And for the condition of the aging time increasing, the ignition time decreases from 57s to 32s, and the heat release rate will increase from 22 kW/m² to 85 kW/m² in which the ignition and fire trend of the cable insulation sheath increase.

An intelligent home energy management system was proposed. Reinforcement learning and a markov prediction model were used to help the system make decisions. The Markov model predicted the future state of users or the weather, and the intelligent decisionmaking support system sent signals to local controllers to control furniture. This work benefits energy management because if the system knows the user's next state, it can control a specific appliance to save energy. Meanwhile, if the system can predict the weather, the house can use green energy rationally. The proposed energy management system could be applied in an intelligent house, city energy management systems, and building energy management. The state prediction helped the decision-making system make accurate and rational decisions.

In order to reduce the power outage loss of the grid after a large area power failure in the regional power system, it is necessary to start the black power grid as soon as possible. For distribution networks with distributed generation (DG), it is of great significance to determine the partition scheme of islands and put forward the black start strategy of DG to continue to supply power to the loads, so as to reduce the power outage loss. Fully considering the importance of loads and their controllability, an island partition scheme of two steps are proposed based on the established mathematical model. The primary island is formed by using the dynamic programming algorithm, and then, the secondary island is formed by modifying primary island according to certain rules, thus obtaining the optimal island. Then the starting sequence of the black start power supply in the island is determined by Analytic Hierarchy Process (AHP) and the black start strategy is obtained. The effectiveness of the proposed island partition and black start strategy is verified by using the example of PG&E69-node system.

Through the impact of fully mechanized coal mining face security personnel, equipment, operating environment and the analysis of geological environment, the establishment of a fully mechanized coal mining face safety evaluation index system is presented based on rough sets - Support Vector Machine fully mechanized coal mining face Safety assessment model. Model application method of rough set evaluation index system of attributes, eliminate redundant attributes, thus reducing the input of SVM learning sample dimension. On this basis, the use of SVM regression nonlinear characteristics was evaluated, and the final example shows the applicability of this model.

Voice assistants, such as Amazon Alexa, Apple Siri and Tmall Genie, using voice biometrics for the identity authentication, are becoming pervasive in our daily lives. However, voice assistants are vulnerable to reply attack due to the open nature of voice-input channels. An attacker can record the voice commands of victims and replay them to spoof voice assistants. Existing liveness detection approaches are mostly based on machine learning methods, which are expensive and complex. Recently, several approaches are proposed to leverage the human specific voice features or the distinctness of voice played by loudspeaker. However, they require the users and the voice assistant to be in a fixed position and at a very close distance, which is not user-friendly in practice. This paper proposes LiveEar, an efficient and easy-to-use liveness detection system for voice assistant. LiveEar utilizes the differences in phoneme positions between live-human voices and voices replayed through loudspeakers. Specifically, it calculates the time-difference-of-arrival (TDoA) in a sequence of phoneme sounds to the microphone on the voice assistant. Then, an SVM-based classification model is trained with the extracted TDoA features. This paper implements a prototype of LiveEar and evaluates its performance using real-world data. Results show that LiveEar achieves high detection accuracy in various flexible positions, with negligible runtime overhead.

Tone mapping operators (TMOs) and multi-exposure fusion algorithms (MEFs) are employed to visualize the informative contents of high dynamic range (HDR) images on the standard dynamic range (SDR) devices, but also lead to visual quality degradation. Thus, it is urgent to evaluate the perceptual quality of tone-mapped images (TMIs) and multi-exposure fusion images (MEFIs). This paper proposes a blind quality index for TMIs/MEFIs by capturing exposure, colorfulness and structure characteristic features. The connection between feature space and associated subjective ratings is established via a regression model. The proposed method is tested on the public database and compared to the state-of-the-art blind image quality assessment (IQA) methods. The experimental results show that the proposed method achieves superior performance.

This paper proposes a sub-channel image reconstruction algorithm based on Delaunay triangle. First, according to the different sensitivity of the human eye to the three color channels, different channels are used to avoid the influence of watermark embedding on pixels around the feature points. Then, feature points should be selected on the selected channel to make them have sufficient robustness. At this time, the location of the feature points extracted on the corresponding channel can be used as the watermark embedding position. Finally, considering the influence of geometric attacks such as affine transformation, this paper proposes a reconstruction algorithm based on Delaunay triangle, which uses feature point information to reconstruct the image before the watermark extraction, and extract the watermark on the reconstructed image. The effect of watermark extraction has been improved.

Quantifying the pre-trained neural network model can reduce its storage size and speed up the forward-inference process. Previous researches for the quantization of voice wake up networks couldn't reach our anticipation for applying it to applications. The voice wake up network is mainly composed of recurrent neural network and classification function. Recurrent neural network can take preorder words into consideration and it widely used in voice network. In this paper, we propose method to quantize the structure of recurrent neural networ including RNN, GRU and LSTM by optimizing the method of quantifying the activation function. At the same time, using the translation invariance of the classification function to improve classification accuracy on Speech commands dataset version 2. Using the method proposed in this paper, we can make the quantized model achieve a better voice wake up rate.

At present, as the knowledge graph continues to increase, problems such as incomplete information in the graph also emerge, mainly reflected in the lack of links between entities [1], and the current commonly solution is to learn their representations that can represent more semantic information for the entities and relations in the graph. This article focuses on the temporal knowledge graph (TKG), consistent with the goal in the static knowledge graph, this paper tries to propose a better representation learning method to solve the link prediction problem. According to the time information in the temporal knowledge graph, the paper constructs a bag of words about time, uses the bag of words model to decompose the time in the temporal knowledge graph into sequences, and learns its semantic information together with relations and entities. The embeddings obtained in this way can learn more semantic information. In addition, proving the effectiveness of the model through experiments on the currently commonly used temporal knowledge graphs.

Knowledge graph is a kind of typical artificial intelligence technology, which can improve cognitive ability of machines. Space is a booming domain for human beings. In this paper, space knowledge graph is proposed in order to improve intelligence of space information system. A major obstacle of space knowledge graph construction is to identify and extract core named entities and relations in space domain. An unsupervised learning method to extract core named entities and relations is proposed in this paper. Firstly, confidence level index used to measure importance of entities and conditional confidence level index used to measure credibility of relations between entities are defined and investigated. Then, an entity and relation extraction procedure based on maximum likelihood estimation and apriori algorithm is designed. Finally, a case study is carried out and influence of super parameter setting on the extraction result is discussed. The method proposed in this paper can be used to construct space knowledge graph at a lower cost compared to supervised method, and can be extended to other domain knowledge graph construction.

Thermoelectric modules are widely studied by domestic and foreign experts and scholars because of their light weight, small size, absence of mechanical moving parts, and the ability to switch between cooling and heating and adjust the cooling/heating capacity by current. Based on the advantages of thermoelectric modules, this paper proposes an innovative thermoelectric module-based intelligent temperature control device for instantaneous high-power payloads of spacecrafts, which can realize the rapid heat dissipation of high power payloads and adjust the current of thermoelectric modules through intelligent temperature control strategies to precisely regulate the payload temperature. The results show that the thermal control device can efficiently control the heat of the instantaneous high power load during the orbit, and the precision temperature control of the thermoelectric cooling work during the load operation is achieved with the temperature control accuracy of ±0.5°C.

Fault information in analog circuits is complex and diverse, so as to improve the accuracy of fault diagnosis, a Kent mapping and Particle Swarm Optimization (PSO) combined optimization Extreme Learning Machine (ELM) model is proposed. Firstly, the original data set of the circuit is normalized to obtain the fault data set. Secondly, Kent mapping is used to initialize the position of particles in the particle swarm, which makes the initial particle swarm more evenly distributed in the search space and enhances the global optimization ability. Third, aiming at the problem of the input weight and hidden layer bias generated randomly by the ELM are easy to lead to poor generalization ability, the Kent-PSO algorithm is used to optimize the input weight and hidden layer bias of ELM to obtain better and more stable ELM network parameters and improve the fault diagnosis ability. The diagnosis example of Sallen-Key bandpass filter shows that the proposed method has better fault diagnosis performance than PSO-ELM model.

In order to solve the problem of inaccurate channel estimation of least square (LS) algorithm for multiple input multiple output orthogonal frequency division multiplexing (MIMO-OFDM) system under time-varying channel, a novel channel estimation method using exponential smoothing algorithm (ESA) is proposed in this paper. Based on LS estimation algorithm and ESA theory, the proposed method fully exploits the variation law of time-varying channel, adopts the predicted results of the previous time to predict the estimation value of the current time by iterative computations, and utilizes the low-pass character of the ESA equivalent filter to reduce the random disturbance. Therefore, the additive white Gaussian noise (AWGN) existing in LS algorithm is effectively suppressed and the estimation accuracy is further improved. Simulation results show that, compared with the traditional channel estimation approaches, the proposed method can provide high estimation accuracy with low computational complexity, especially under time-varying channel and low SNR conditions..

Feature matching is still a crucial and challenging problem in image processing. In this paper, a novel feature matching algorithm based on local relative velocity consensus (LRVC) is proposed to find two accurate matched feature points. To remove the outliers accurately and robustly, the relative velocity between the putative matches and their corresponding neighbors are exploited. Based on the consensus of relative velocity and the consensus of neighborhood elements, the similarity of each putative matches is evaluated. With a two iteration outlier removal strategy, the feature points are matched accurately and robustly. In the experiment, VGG dataset are used to verify the accuracy and robustness of the proposed method LRVC. The matching results indicate that our method is superior to three other classic feature matching methods.

A 3dB power divider is designed based on E-plane waveguide directional coupler and self-compensating waveguide phase shifter. This power divider was implemented with a center frequency at 110GHz. Simulation results show that amplitude balance within ±0.3dB and phase balance within +0.5/-1.2 degrees, and isolation is better than -15 dB from 100 to 120 GHz. It has good performance in-band flatness and bandwidth is 18%. It can be used in terahertz communication system.

Internet data center module(IDC) is an important part of the multi-station integrated substation, and its site selection will be an important basis for the site selection and construction of the multi-station integrated substation. In order to be useful to the practical engineering, the fuzzy analytic hierarchy method is proposed to analyze the location of IDC. There are six factors influenced in total which land price, policy environment, the annual average temperature, industrial electricity, network environment and peer development. FAHP has been adopted for calculating and sequencing priority membership value for every single factor thoroughly. According to the factor, different priority location has been chosen. Finally, based on the fuzzy analytic hierarchy process, the paper proves the effectiveness and convenience of the site selection method, through analysis of the three regions.

In the process of spacecraft thermal test, the infrared cage is mainly used to simulate the heat flow outside the space. In the design, the infrared cage is divided into several heating zones corresponding to the spacecraft. The heating strip of the infrared cage is divided into heating circuits, the heating circuits are grouped and collected, and the infrared cage heating node table is defined as an important basis for controlling the infrared cage. At present, there is a single method of grouping and gathering heating circuits, which is mostly limited by manual adjustment based on experience and lack of automation. In order to solve the shortcomings of the existing technology, this paper adopts the genetic algorithm hybrid BFD based on reinforcement learning to group and collect the heating circuits, and optimize the design results according to the constraints. The user interaction interface is designed to facilitate the user to do further analysis. Thereby improving the level of business automation, optimizing working procedures and reducing costs.

In some electronic reconnaissance fields, fast frequency switching is required. A radio frequency signal generator with fast lock and accurate lock time appears to be of great significance. Traditional detection requires additional equipment, which is expensive and cumbersome to operate without portability. This paper designs a simple, portable and easy-to-operate lock detection method, uses a timer to monitor the LD pin of the phase lock chip in real time, optimizes the algorithm for capturing signal edge transitions, and calculates the frequency switching time. The ADF4351 fast lock mode is combined with lock detection to improve the topology of the loop filter and speed up the lock time. With the keyboard and LCD module, a radio frequency signal source system capable of self-checking and locking time is constructed. Simulation and actual results verify that after changing the filter structure to achieve quick lock, the lock time is optimized by 16%, which verifies the effectiveness of quick lock and the accuracy of lock detection, and realizes efficient and convenient lock time detection.

In the transmission line inspection, in order to obtain wide-view and large-scale images, it is necessary to stitch images taken by drones. This article proposes an improved SIFT-LATCH-based UAV aerial transmission line image stitching algorithm. First, the image is processed in parallel through multi-threading. The SIFT algorithm is used to extract the feature points in the image, and the LATCH descriptor is used to describe the feature points. The sampling consensus algorithm is used to purify the feature points, and finally the translational homography matrix is used to solve the problem of black lines in the stitched image. Experiments show that the algorithm proposed in the article has the advantages of good real-time and high accuracy in splicing UAV aerial pictures of transmission lines.

According to the linear quadratic optimal control principle, the idea of optimal tracking control is applied to the three-phase inverter control system, which can effectively improve the output characteristics of the system. Based on the establishment of mathematical model, the nonlinear load and load mutation are regarded as the uncertain factors affecting the performance of three-phase inverter, and the parameter optimization design of state feedback and input feedforward is carried out based on the linear quadratic optimal control method. The simulation results show that the LQR optimal tracking control can better adapt to the adverse effects of nonlinear load and load mutation than the traditional quasi-PR control, and the system has good harmonic performance, dynamic performance and the ability to resist load disturbance.

This paper presented a 3D model identification method by matching pairs of rendered images and depth images. Firstly, the technique uses the HARRIS detector and PIIFD detector (partial intensity invariant feature descriptor) to detect feature points of rendered-depth image pair of shape and then uses LGHD (Log-Gabor Histogram Descriptor) to descript these feature points. Secondly, the shape identification processing is conducted, including normalizing shape pose, capturing multi-view pairs of rendered images and depth images, and image matching. Finally, 100 pairs of 3D models are used for the experiment. The experiment results show that the proposed method has an efficient performance on rendered-depth image matching and can be used for 3D model identification.

The speech denoising model based on adversarial generative network has achieved better results than the traditional machine learning model. In this paper, for the short cut connection in the generator, we discuss its influence on the information transfer between encoder and decoder, and propose SDGAN at target. SDGAN sets linear and convolution filters in the short cut connection which adaptively learn the optimal information processing. The information filter still enables the generator to solve the gradient vanishing problem, and it can also avoid information redundancy and improve expression ability. In addition, SDGAN replaces the L1 regularization term in loss function with the L2 regularization term, which not only makes the output speech of the generator closer to the clean speech, but also avoids sparsity. In the experiments, SDGAN significantly performs better than other traditional GAN in five performance metrics (such as PESQ), and the effect of convolution filter is better than that of linear filter.

Internet has become an important information platform, and it is very important to accurately understand the multimedia information of the Internet. In this paper, our main task is to do classification based on pictures and texts collected from the Internet, which is a classification problem of multimodal fusion in practice. However, when multimodal data is put together, there may occur the dimension disaster problem. We apply feature selection (FS) and dimension reduction (DR) in feature levels both in later fusion and early fusion to solve this problem. The classification accuracies in different models obtain improvements in different levels respectively. We also discuss the relation between single modals and multimodal in later fusion. In our experiments, images and text can be classified by multimodal models under FS/DR, and of which with the help the multimedia information from the Internet can be analysed better to help enterprises provide better services and products, and then carry out better network marketing and promotion.

This paper takes the electronic fence technology of power grid infrastructure project based on Beidou Positioning Technology as the research object, summarizes the use situation and characteristics of this type of electronic fence technology, carries out technical and economic comparison, comprehensively considers the different use needs in the whole life cycle process of substation project infrastructure construction, operation and maintenance, expansion and technological transformation, and considers the purchase cost, installation difficulty, etc. This paper summarizes the technical advantages and engineering pilot application experience of this type of electronic fence technology. Taking one 220kV substation construction project as an example, this paper comprehensively considers other types of electronic fence technology, carries out technical analysis and economic comparison, and provides guidance for the deepening application of electronic fence technology in power grid infrastructure project based on Beidou Positioning Technology in the future.

When the load current check point and compensation injection point of the Static Synchronous Compensator (STATCOM) are at different positions and there are differences in phase and amplitude, there will be a detection difference. To solve this problem, this paper proposes an improved STATCOM compensation command current detection method based on the ip-iq algorithm. This method can eliminate errors and improve detection accuracy by setting the preset phase angle and proportional integral (PI) controller according to the voltage phase and amplitude difference between the check point and the compensation point, thus breaking through the traditional ip-iq. The algorithm is only applicable to the situation where the detection point and the compensation point are at the same position, which improves the accuracy of detection. Theoretical analysis and simulation results show the effectiveness of the method. S

Since the electrical equipment with nonlinear load began to be used, the harmonic suppression technology has become the focus of research. In this paper, a robust multi resonance servo control method is proposed and applied to the medium frequency power supply. The method can not only perform the closed-loop system stability and asymptotic tracking, but also make the power supply output possess the advantage of rapid dynamic response, and low-order strong harmonic suppression. First, this paper presents a mathematical model, analyzes the time-domain robust control - linear matrix inequality and regional D stability and turns the analysis of the problem of robust control problem to a one of linear matrix inequality (LMI) convex optimization problem through the augmentation of system matrix dimension. Then, the YALMIP toolbox is used to solve for the status feedback gain matrix. Finally the whole system is simulated and tested. The results show that the designed controller makes the dynamic recovery time of the closed-loop system less than 5ms, and the output voltage THD is reduced to less than 4.5% with the nonlinear load.

In multi-camera pedestrian tracking, pedestrian apparent features are usually used to solve the problem of cross-camera reidentification. In order to make the pedestrian apparent features robust to the directional changes of pedestrians in different cameras. In this paper, a pedestrian feature extraction method combining directional features is proposed by using deep network Resnet 50. Pedestrian feature extraction network is trained by adding pedestrian direction information. in order to improve the performance of person reidentification model, BNblock structure is added in the design, a batch normalization (BN) layer is added after the feature to get the normalized feature. It can make the triplet loss converge with the convergence of the ID loss, thereby improving the performance of the model. The proposed method was verified on DukeMTMC and Market-1501 data sets, and the results show that the pedestrian appearance features combined with direction information proposed by the pedestrian reidentification algorithm can significantly improve the reidentification accuracy. By adding BNblock into the network structure, the accuracy can be further improved.

The phase of the object contains the depth, shape, refractive index and other information of the object surface, which is more important than the intensity. As a classical phase retrieval algorithm, the Transport of Intensity Equation(TIE) can directly obtain the phase calculated from the known intensity. In this paper, a new algorithm is proposed to reconstruct the three-dimensional phase information of an object by tomography. The algorithm obtains high-precision phase input through the high-order TIE, and then reconstructs the three-dimensional phase information of the object by using Fourier slice theorem backprojection tomography. The experimental results show that the algorithm can solve the problem of low phase accuracy caused by intensity differential approximation constraints, and can obtain high-precision 3D phase reconstruction results of objects.

An analytical method for electromagnetic scattering characteristics of complex satellite targets by using multilevel fast multipole algorithm (MLFMA) and short-time fourier transformation (STFT) is proposed in this paper. By constructing different electromagnetic models of satellite target with multi-scale structures and coatings, the aspect and polarization characteristics of radar cross section (RCS) and time-frequency representation (TFR) are studied. To validate the influence of the multi-scale structure and coating, the full-polarization RCS at specific angles, RCS statistical characteristics and TFR correlation are simulated and compared. It is concluded that the multi-scale structures and coating materials should be sufficiently considered to calculate the accurate scattering characteristics.

The classification of superpixel images by graph neural networks has gradually become a research hotspot. It is a crucial issue to embed super-pixel images from lowdimensional to high-dimensional so as to turn complex image information into graph signals. This paper proposes a method for image classification using a graph neural network (GNN) model. We convert the input image into a region adjacency graph (RAG) composed of superpixels as nodes, and use residual and concat structure to extract deep features. Finally, the loss function that increases the distance between classes and compactness within classes is used as supervision. Experiments have been tested with different numbers of superpixels on multiple datasets, and the results show that our method has a great performance in superpixel images classification.

In the information age, with the development of internet technology, there are more and more various business services, at the same time, network security issues are also increasing. However, traditional protection methods are gradually hard to cope with the complex network environment. Aiming at the credible connection establishment and data communication problems that need to be solved urgently in the credible network, an interentity communication mechanism based on end-to-end credibility assessment is proposed. This mechanism combines trusted computing with network access control mechanisms, integrates network access authority control technology, server identity authentication technology, communication subject measurement technology and network link transparent encryption technology, adopts real-time monitoring of terminal status, dynamic adjustment of access authorities, optimized network access control mechanism and data transmission method, to against security issues of the network communication. After security analysis, this mechanism improves the security of the connection and communication between the terminal and the network. It can effectively avoid security threats to information systems, and make the service of the information system safe and reliable.

Internet of Things devices are faced with ever-increasing amounts of data, and they can no longer only do data collection as they did in the past, and hand over computing tasks to servers on the cloud. The growing computing requirements in the field of Internet of Things and the diversity of its scenarios put forward configurable and flexible customization requirements for customized processors. This paper conducts in-depth analysis and research on configurable customized processors, and analyzes related work in the field of RISC-V chips and deep convolutional neural networks in the field of device-side optimization at home and abroad. Based on this, this paper proposes an instruction and hardware design of a customizable deep convolutional neural network accelerator attached to Rocket-Chip Generator based on the RISC-V modular instruction set.

In the field of video surveillance, with the continuous development of the Internet and artificial intelligence, the related technology for single device has been very mature, but how to accurately realize pedestrian tracing under cross-devices is still an active research field. In this paper, a multi-granularity pedestrian traceability method is proposed, which is based on the attribute extraction of pedestrian image and the person re-identification algorithm based on attention mechanism retrieves the same pedestrian image cross a lot of devices. The results show that the video surveillance system has good performance, which reduces the occurrence of tracking loss, and the track is visualized at the platform.

Traditional fruit and vegetable classification is mostly based on manual operation, which is inefficient. Deep convolution neural network shows excellent performance in feature learning and expression. In this paper, an automatic recognition system of fruits and vegetables based on deep convolution neural network is designed. By using depthwise separable convolution instead of the traditional standard convolution, a neural network is constructed with less parameters, which is suitable for equipment with limited resources. A small data set including 12 kinds of common fruits and 8 kinds of common vegetables is established for training and testing through network download and physical shooting. The experimental results show that the recognition accuracy reaches 95.67%.

Nowadays, with the rapid development of urbanization worldwide, the indoor area has become complex. Even with floor maps and guidance signs, people waste time on self-location in an unfamiliar building for the reason that the mature location services like GPS are limited as outdoor positioning series, and the method of indoor positioning is relatively missing or rough. Therefore, the better indoor positioning systems are in a highly command. The indoor positioning system has the goals for realizing the location marked and tracking of people and objects in indoor scenes. This paper focused on the main positioning method in use, which is distance-based positioning algorithm. Then, the advantages and disadvantages of the mainstream algorithms RSSI and AOA were discussed. At last, by comparing the current research, the author put forward an optimized indoor positioning algorithm based on AOA, which may become the mainstream in the future.

How to predict a more accurate next frame according to the current frame is still a problem. In this study, a method of generating intermediate frames in VR is proposed, which can effectively reduce the image jitter in VR games and increase the user experience; on the other hand, the inverse matrix method proposed in this study has a small amount of calculation and a fast calculation speed, and can complete the calculation in a short time. Therefore, this method can help us to complete the insertion of frames in the specified time. In this paper, firstly, the algorithm is constructed based on the principle of inverse matrix, and then the simulation experiment is carried out. The experimental results show that the scheme is a viable and effective solution.

Real time, fast and batch processing of vibration signals has become a future development trend in the field of fault diagnosis, but data dimensionality disasters may arise. In view of the long running time of deep learning in the case of large samples, the gradient descent method and its variant algorithms are introduced for the loss function optimization problem, and the approximate optimal solution is solved in an iterative manner. The gradient descent method is used to minimize the loss function, and the research is carried out on the basis of MATLAB program implementation. The gradient descent method and its variant algorithm are applied to the rolling bearing fault diagnosis model for analysis. By comparing the algorithm's convergence speed, loss value and accuracy of the rolling bearing fault diagnosis model, a relatively good optimization algorithm suitable for the rolling bearing fault diagnosis model is determined.

Aiming at radar main lobe false target jamming, the forms of target echo and jamming signal are analyzed, and its modeling and simulation are carried out. By extracting the 11 dimensional features of time domain, frequency domain and amplitude fluctuation characteristics, it is proved that it is distinguishable for real target echo and false target jamming signal through simulation experiment, and a feature model is established; aiming at the shortcomings of traditional classification model, bagging algorithm is applied to deal with the over fitting phenomenon. Simulation results show that the feature model established in this paper has good performance in bagging algorithm.

When a single-phase-to-ground fault occurs in the distribution network, there is a wealth of information in the transient process of the fault line, from which the fault line selection criterion can be extracted. In this paper, based on the analysis of the related principle of wavelet transform and the concept of entropy, the comprehensive wavelet entropy is defined, and the principle and process of fault line selection are put forward by analyzing the variation coefficients of a group of comprehensive wavelet entropy. Finally, the simulation software is used to verify the fault line selection method under different fault conditions to prove its applicability and feasibility.

Dampers can prevent wires from breaking due to vibration, so it plays an important role in ensuring the stable operation of transmission lines. But the accuracy of popular object detection algorithm is still very low, because the occlusion phenomenon of dampers is serious due to the variable shooting angle. In this paper, we introduce our Supervised Attention RCNN (SA RCNN) to improve the detection effect of occluded dampers. SA RCNN mainly consists of two components: Box Supervised Attention Module (BSAM) and One Proposal Multiple Predictions (OPMP). In order to solve the problem of false detection caused by similarity of features, BSAM enhances features by using ground truths as supervised information to guide the generation of attention. In order to solve the problem of missed detection due to the NMS strategy, OPMP predicts a set of instances for one proposal rather than a single one. Experiments show that SA RCNN obtains 4.2% AP improvements on the dataset of occluded dampers compared to FPN.

Since entering the information age, the popularity of mobile communication has become higher and higher, and people's requirements for communication quality are also increasing, but the channel resources that people can control are limited. This paper proposes an optimization algorithm for channel resource allocation based on water-filling algorithm. Some improvements have been made on the basis of the original water-filling algorithm. The sgn function is used to optimize the transmit interference temperature limit (Sk), so that the screening result is not completely dependent on Sk, which improves the flexibility of channel screening and effectively optimizes the capacity of the communication network.

Aiming at the problem that visual SLAM (Simultaneous Localization and Mapping) cannot satisfy both high-precision positioning and rapid mapping in large scenes, this paper proposes a map construction method. First, when calculating the pose, the method of combining EPNP and graph optimization is used to optimize the pose estimation to improve the accuracy of the pose. Then, the paper uses Lie algebra to define the vertices and edges of the pose graph, which improves the convergence speed of the model. Finally, when constructing a three-dimensional map, a divide-and-conquer map method is proposed, which uses multiple sub-maps that contain each other for point cloud registration and down-sampling filtering, so as to improve the accuracy of map construction. Experimental results show that the algorithm in this paper has good performance. In the TUM test set, the speed can reach 0. 139s/frame, which is about 0.2s higher than the original method, and the accuracy error RMSE value is reduced to 0.055m. The feasibility of the algorithm is proved, and it can meet the accuracy and speed requirements of 3D dense map construction in large scenes.

With the rapid development of power industry, the number of substations is also increasing, and the voltage level is also higher and higher, and the types of equipment are diverse. The detection of equipment fault signal in substation is an important part to ensure the normal operation of electric power. Only timely detection of fault signal and corresponding processing can ensure the overall operation of substation. But the fault signal often contains a lot of white noise, which will interfere with our judgment of the fault point. How to extract the effective signal from the noisy signal becomes particularly important, but under different signal-to-noise ratio, the selection of different wavelet basis and threshold estimation also have a great impact on the noise reduction effect. In this paper, we choose different wavelet bases and threshold estimation under several different signal-to-noise ratios to study their influence on the effect of signal denoising.

The complex structure of interconnected power system and the existence of external load disturbance make it difficult to establish accurate system model and realize stability control with preset performance. In this paper, a fuzzy adaptive controller design method with preset performance is proposed for uncertain interconnected power systems. Firstly, the system approximation model is obtained by using the fuzzy approximation principle, then the preset performance control strategy is introduced to transform the steady-state preset performance of the system into an error performance function. Finally, the adaptive controller design with preset performance is realized based on the Backstepping method. For the validity of the algorithm, the Simulink simulation is carried out. The results show that the controller can make the output of the system uniformly bounded, that is, the tracking error can always be controlled within the specified performance range.

While great progress has been made in face detection, one of the remaining challenges is to detect the faces of small targets in images. The small target face we define, mainly includes the absolute size of the face is not greater than 32×32 pixels and the relative proportion of the face size is not greater than one tenth of the image size. To meet this challenge, we propose an effective face detector, called SmallFaceBoxes, which has a better performance in both speed and accuracy. Specifically, our method has a lightweight but powerful network structure consisting of small target face detection layers and multi-scale convolution layers. The design of small target face detection layers enables detector to increase the representation ability of small and medium-sized faces in the image and improve the detection accuracy of small faces. The multi-scale convolution layers aim at enriching the receptive fields and discretizing anchors over different layers to handle faces of various scales. In addition, according to the size of receptive field, we propose a strategy of setting Anchor in multiple feature layers, thus improving the recall rate of small face. As a consequence, our algorithm performs well on the wideface dataset.

To improve the survivability and penetration probability of unmanned undersea vehicle(UUV) on near sea bottom combat missions, the factors leading to sonar detection blind zone are analyzed, and the detection blind zone produced by underwater terrain obscuration is mainly studied. A detection range calculation method based on DEM and acoustic ray is proposed, and the simulation study of the near sea bottom penetration path planning is carried out by using terrain blind zone and invasive weed optimization algorithm. Simulation results show the effectiveness of the proposed algorithm. This method can be applied to path planning of the UUV on near sea bottom penetration, which helps to improve Combat effectiveness and Operational effectiveness of UUV.

In recent years, hash is a popular method of image retrieval. Convolutional neural network is used to generate image features and hash codes, so as to achieve fast and effective image retrieval. In order to solve the problem of poor image retrieval effect caused by complex background of Multi-target Image, this paper proposes a hash generation method based on the combination of target region recommendation network and convolutional neural network. The RPN network is selected as the target region recommendation algorithm. After the image passes through the RPN network, multiple target regions and the four-dimensional coordinates of each region will be generated. According to the four-dimensional coordinates, the main target region will be filtered and extracted. Then, the feature of the largest target area is extracted by the convolutional neural network (GoogleNet), and the corresponding hash code is generated, which is finally retrieved in the database. Experiments are carried out on voc2012 data set and self collected data set to verify the algorithm. When the number of test images is 1000, the experimental results show that the total correct rate of the retrieval results of this method is 95.5%, which is about 5 percentage points higher than the existing methods.

A three variable lens control circuit based on C8051F311 is designed to solve some problems at the scene such as low precision or difficult quality control. In this system, the C8051F311 is used as the control, first a PWM signal is output, then the PWM signal through a RC filter circuit, finally through a signal processing circuit composed of comparator chip TL064 and amplifier IC TDA2030A to control the brushless DC motor, based on above all, the control of the lens is realized. At the same time, the circuit parameters are designed by using Multisim to simulate the circuit. The practice shows that the system has simple structure, convenient speed regulation and stable operation.

When multiple photovoltaic power sources are connected in parallel in the DC distribution network, photovoltaic with different output have differences in the regulation capacity of unbalanced power of the system. In order to make full use of the power margin of photovoltaic to adjust the DC bus voltage, this paper proposes an adaptive droop control of DC distribution network considering the photovoltaic output power margin. The voltage signals of the input and output terminals of the PV side converter are used to judge the system load change and the output of the photovoltaic power supply. On the basis of traditional droop control, the droop coefficient is modified based on the input voltage deviation of PV side converter to realize the reasonable distribution of the unbalanced power between photovoltaic power sources with different power margin. The output impedance characteristic model of PV side converter is established, and the adaptive droop control proposed in this paper can effectively improve the stability of multi PV parallel connected DC distribution system. Finally, the effectiveness of the proposed control method is verified by MATLAB/simulink platform simulation.

Multi-paragraph aims to allow machines to read multiple paragraphs and infer the answer to the given question. Usually, the model needs to use the selector to narrow the range of candidate paragraphs, and then use the reader to find the answer. For the selector, the previous work generally matches the question and the paragraph on the word level and the paragraph level. In this paper, we study the encoding and matching algorithm at the chunk level, which is supposed to make our model more accurate. Combining it with gated self matching mechanism, we design a multi-granularity matching network as the selector. The experimental results show that our model achieves competitive performance on Quasar-T.

In this paper, a novel noise level estimation technique is presented based on skewness fitting. The difference between exact skewness and observed value on each channel contributes to an objective function. The skewness concentration is helpful to guarantee the reasonability of the model. The optimal solution means an estimation of the noise variance and it is easy to be applied in some parametric denoising model. The accuracy and efficiency of proposed algorithm were verified by the numerical experiments.

Aimed at solving the problem of the end shield fault of 500 kV capacitive bushing which happened frequently in the main power grid in recent time, two faults are taken as examples to analyze the fault characteristics and reasons of the spring cap type end shield devices. On that basis, effective revention measures and modification schemes for the spring cap type end shield devices of 500 kV capacitive bushing are proposed from perspectives of the equipment security operation and maintenance, and personal safety.

In order to find the best path in wireless sensor networks, a dynamic energy-saving routing strategy based on multi hop transmission is proposed. The dynamic state transition optimization rules are designed to increase the search probability of new nodes reasonably, so as to find the global optimal solution quickly and effectively, so as to save the search time, increase the optimal path search probability and prolong the network survival time. Simulation results and analysis show that the dynamic route selection method of D2D communication under multi hop transmission greatly increases the search probability of the global optimal solution, achieves the global optimal solution quickly and effectively, saves the energy consumption of nodes, and is conducive to extending the network survival time.

Machine translation is an important task in natural language processing, and the study of Tibetan-Chinese neural machine translation is of profound significance in promoting Tibetan-Chinese scientific and cultural exchanges and the development of education and culture. In this paper, we investigate the performance of these techniques and methods on Tibetan-Chinese NMT with few samples by using deactivated word lists, data augmentation (back translation), pre-training models (ELMO), and attention mechanisms for the techniques and methods widely used in NMT, using seq2seq and Transformer models as the baseline, and finally, the BLEU value of Tibetan-Chinese NMT is increased from the initial 5.53 to 19.03.

Wikipedia has set up a manual of style to provide the Wikipedian with guidelines to create links as spotlights in articles to help readers' understanding. Wikification is coined following Wikipedia's style guide as a task of automatically detecting mentions and linking to entities in plain-text articles. However, existing methods rely heavily on the mention-entity dictionary and focus on in-map entity precision. Some popular methods are limited to Named Entities. The comparison on the outcome of spotlight annotation is insufficient. We investigate the design challenges for constructing automatic annotating in Wikipedia-style systems. A dataset is built simulating real-world spotlight annotation problems in both the preparing and application phases. Popular methods are compared along with a straightforward framework we propose called WikiER. Experiments show that the performance of currently available systems varies a lot. It is important to train on the task-specific data. The best performance is achieved by WikiER.

Aiming at the problems of high dimension and sparse data in traditional text classification methods, this paper proposes an e-mail classification method based on i-cnn model by combining convolutional neural network (CNN) and inception V1 model. In the convolution and pooling operation, 1 * 1 convolution kernel is added to reduce the thickness of eigenvectors, reduce the parameters and improve the computational performance. Through data validation, i-cnn model for e-mail classification results as high as 92.18%, in the comparative experiment, i-cnn model compared with several machine learning classification models, achieved the highest classification accuracy, in the comparison with or without the emergence structure model, i-cnn model accuracy is higher than CNN model. It shows that the model has a good classification effect, and the integration of inception V1 model can improve the accuracy of text classification.

In this paper, we study a saddle point problem and propose an inexact double primal-dual algorithm for solving the saddle point problem. The sequence generated by the proposed algorithm is proved to globally converge to the saddle point of the saddle point problem in the ergodic sense, and the convergence rate of the algorithm is also obtained. For the model of total variation image processing, the performance of the algorithm is demonstrated. Simulation results show that our method is more efficient in image denoising compared with existing methods.

In order to achieve the purpose of early diagnosis of Alzheimer's disease, this study uses machine learning technology to process 14,659 original data from the public data set ADNI. First use the KNN algorithm to fill in the null value. Then a feature transformation method is proposed to transform regression problems into classification problems. Then this study calculates the trust level of each sample based on the feature correlation and the weight of null value, and proposes an improved LightGBM algorithm-Trusted-LGBM. Finally, the prediction results of the Trusted-LGBM algorithm are compared with SVM, XGBoost, and the unimproved LightGBM algorithm. The experimental results show that the Trusted-LGBM algorithm proposed by this research has a higher Fl-score (0.784) and AUC (0.91). It can be seen that the method proposed in this study can more effectively support the early diagnosis of Alzheimer's disease.

Robotic manipulation for rigid objects is a relatively easy task, while grasping highly deformable objects, such as garments, is still a big challenge for robots. This research detects the grasping points of hanging garment using deep learning to facilitate robot manipulation. A neural network is proposed to predict the cartesian coordinates and visibility of predefined grasping points. In order to reduce the impact of different clothing colors, depth images are used for the input of the model. It is inaccurate and unrealistic to label data manually because of complex dynamics of clothes; therefore, synthetic dataset is leveraged to train the neural network. This paper makes use of generative adversarial network (GAN) to translate synthetic data to real data. A real dataset is acquired taken by Azure Kinect sensor as the test dataset. The experimental results indicate that our method can provide accurate prediction of the grasping points and can be applied to real scenarios.

Mobile manipulators (MMs) performing the tasks of autonomous navigation, object grasp and transportation are widely used in industrial production and other domains. This paper presents a control system with multiple functions and good extendibility for a commonly configured MM, which controls the multi-degrees of freedom (DOF) movement of the MM while navigating, positioning, pose estimating and object grasp. This system has shown desirable control performance on an experimental platform.

The verification of security protocol implementations is notoriously difficult and important. In this paper, combining with the model learning using Tamarin, a formal verification tool of protocol specification, a formal verification method for security protocol implementations is proposed. We extract state machine information from protocol implementations by model learning, and determine suspicious paths in the finite-state machines by cross-validation between different implementations of the same protocol; and then verify whether the suspicious paths violate the security properties of the protocol using Tamarin. It can be used to detect logical errors in protocol implementations and avoid relying on expert experience to make compliance rules from protocol documents when using model checking tools. The effectiveness of this method is demonstrated by the vulnerability detection of the typical ChangeCipherSpec in the TLS protocol implementation. The method proposed can help developers to develop more robust implementations of security protocols.

Online testing has become essential for the reliable operations of manycore systems using three-dimension (3D) network-on-chip (NoC) fabric. This paper presents a software-based online self-testing solution using bounded model checking (BMC) for 3D NoCs. The proposed method implements BMC on the extended finite state machine of the 3D NoC and extracts leading sequences to triggerinternal functions. Next, it detects structural defects in each function activated by the leader sequence using constrained ATPG. Finally, it effectively organizes packets to test the 3D NoC and reduce test cost. Experimental results demonstrate that the proposed method achieves high fault coverage online with extremely lower test cost and outperforms existing methods.

A closed battery detection system consists of a semiconductor alcohol sensor, a temperature sensor, a STM32F103 microcontroller, and a Windows 10 operating system computer. The ZigBee communication and RS485 communication technology are used to detect the temperature and alcohol concentration of the Dianchi Lake online, which solves the problem of manual sampling and damage to the sealing of the Dianchi Lake. Field tests have shown that the acquisition system data meets the manual physical and chemical test data.

In this article, a refined Lyapunov functional is proposed to study the stabilization problem for Takagi-Sugeno (T-S) fuzzy systems with time delays. Considering both the time-varying delays of the system and the constant delays of the part of sample-data control loop, the refined Lyapunov function in this paper makes it more practical in actual sampling control pattern compared with some existing research. Some techniques of dealing the linear matrix inequalities (LMIs) are used to derive the less conservative criteria and an example is presented in the end to prove the availability of the work.

Tier-to-tier Multi-Shuttle Systems (TT-MSS) become increasingly adopted due to their high flexibility and efficiency. In the Tier-to-tier MSS, shuttles work across different tiers with the help of lifts. How to schedule these devices matters a lot in terms of lowering the expected order cycle time. In this paper, we propose a time sequence task scheduling model considering the movements of shuttles and lifts. The model describes the task scheduling problem between shuttles and lifts, which is generated in a specified time window. An improved genetic algorithm is proposed to solve the objective optimization function in the task scheduling problem. Finally, we illustrate the advantages of the Tier-to-tier MSS by conducting experiments, and the results indicate that with the increase of retrieval tasks, the efficiency of Tier-to-tier MSS is approaching that of TC-MSS with fewer shuttles used.

A kind of intelligent ventilation system is studied in this paper, which includes host system and slave system. The design scheme and schematic diagram of the hardware circuit are given. A variety of sensors are integrated, which can measure temperature, humidity, formaldehyde, PM2.5 dust, smoke and other parameters. According to the measured data, the host sends out control commands such as "open window" " purify air" and "draught fan speed", and realizes intelligent ventilation. The system has dust prevention, damp proofing and other work mode. The power drive circuit designed by SCR designs to realize no spark control. Host monitors multiple slave stations using ZigBee wireless networks, through the WIFI broadband access network and realize the application of "Internet +". Therefore, users can view and monitor the home environment status through the mobile phone APP software. The system can be used in the newly decorated houses without ventilation, and can also be used in various mines, subway stations, working room of chemical plant and other workplaces

In the theory of evidence fusion, it is a very important task to deal with the conflicts between evidences correctly. Based on the credibility degree of evidence proposed in [10], in order to increase the credibility degree of each group of evidence, the concept of falsity is introduced, which is another concept to represent the conflict between two groups of evidence, this treatment makes up for the deficiency of normalized evidence weight coefficient when one group of evidence credibility degree is zero. Because we consider not only the credibility degree of evidence, but also the falsity of evidence, the weight coefficient of unreliable information is effectively reduced. The numerical simulation results show that the algorithm has the advantages of simple idea, fast convergence speed and strong focusing.

Q&A data based on chat records have one characteristic: customers ask, service answers. There is a large amount of knowledge between the questions and answers of customer service. By optimizing the Q&A extraction algorithm to extract knowledge, a very excellent Q&A library can be constructed, thus the accuracy of FAQ system is greatly improved. By analyzing the existing data, this paper cnosiders the extraction strategy of question answering from machine learning and non-machine learning respectively. Then We compares their performance from three aspects of precision, recall and Fl-score according to their different characteristics. In order to ensure the best classification performance, grid search and K-fold crossover are also used to test the optimized classifier performance. After selecting the optimal data extraction strategy, We developed a FAQ system use this strategy, the system results show that the performance is reliable.

In view of the fact that power load forecasting is increasingly becoming an important part of power system planning and power dispatching, this paper proposed a forecasting method based on genetic algorithm optimizing BP neural network and applied it to household power load forecasting. First of all, the original data were pre-processed by factor analysis to eliminate the correlation between variables, and then the household load from January 1st to March 22nd was predicted by using BP and GA-BP models respectively. The author obtained the predicted results in the next 5 days summing to 120 hours and utilized MAPE and RMSE to compare their prediction errors quantitatively. The predicted results of the calculation example are in good agreement with actual values. Therefore it is demonstrated that the BP neural network optimized by genetic algorithm has better forecasting effects, which can provide reference for residential and municipal short-term load forecasting.

The traditional A* algorithm uses Euclidean distance as the heuristic function in the path planning of mobile disinfection robot. This method may lead to the generation of a certain number of useless nodes, which makes the search time longer. To solve this problem, this paper proposes an improved A* algorithm, starting from the heuristic function, adding additional value to the heuristic function, using the cosine value of the angle as the additional parameter. This method effectively reduces the comparison of useless points with the same F value, and significantly reduces the number of visited nodes. The simulation results show that the improved A* algorithm can significantly improve the path planning efficiency of mobile disinfection robot.

In order to alleviate the problem of large-scale grid-connected photovoltaics and increase the pressure of grid peak regulation, this paper proposes a dynamic economic dispatch method considering the system peak regulation margin. The system peak shaving margin is proposed to deal with the flexibility problem in economic dispatch and improve the enthusiasm of thermal power unit peak shaving. The economic dispatch model is transformed into a dynamic programming model and solved by the simulated annealing-Q learning algorithm. The analysis of the results shows that the proposed dynamic programming strategy can effectively improve the operating economy of the system and increase the flexibility of peak shaving of thermal power units.

This paper presents a state-of-the-art algorithm of self-adaptive RBF network PID control and aims at achieving the precise and fast trajectory tracking for a dual-mass servo system. To increase the robustness of servo system parameter varying and external disturbances, a classical PID algorithm with enhanced structure based on RBF neural network is proposed. Extensive simulations show that it practically validates the superiority of the proposed RBF adaptive PID controller. The experiment was simulated respectively under the external disturbance, which indicates that accurate tracking performance of the servo system with dual-mass load has been achieved and also verifies the effectiveness of self-adaptive PID control strategy.

Focused on the problems that the randomness of the noise-adding operation in the de-noising auto-encoder (DAE), and the data matrix does not consider the impact of trusted users on the deep preferences of target users, this paper proposes a recommendation algorithm based on stack de-noising auto-encoder (SDAE) which integrates the preferences of trusted users. Firstly, the score vector is used as the input of the auto-encoder, and the mask vector is designed to train the potential preference of the target user. Secondly, the deep preference of the target user and the trusted user is obtained by the weighted fusion of the features of the two hidden layers of the auto-encoder. Thirdly, in order to reduce the impact of noise on the prediction accuracy, the cascaded auto-encoder model is constructed and trained according to the greedy training method layer by layer training. Finally, SDAE model is compared with other models on different data sets. The experimental results show that SDAE model has better recommendation performance.

P2P network lending is a " peer-to-peer " loan through a third-party Internet platform built by P2P companies. The application of machine learning algorithms to the field of P2P loan default prediction will improve the operating capabilities of the platform, and also effectively regulate the lending market. In this paper, we use Lending Club's loan data and feature engineering technology to apply the XGBoost algorithm to construct a P2P loan default prediction model, and we choose five performances: accuracy, AUC value, error rate, model robustness, and model run time to compare it with Logistic Regression and Decision Trees. The result shows that the prediction accuracy rate of the XGBoost algorithm is 97.705%, which fits the actual results better, and can effectively control the loss cost caused by model errors. In addition, we also select 10 features that have the greatest impact on loan default rates based on the XGBoost algorithm, and provide a reference for P2P lending platforms.

Aiming at the problem of unknown wireless protocol identification methods that focus on the application layer, but the wireless network application layer data is difficult to capture, and the existing method research stays at the signal level and the identification effect is poor, this paper proposes an identification method for feature selection of wireless protocol data from the data link layer. This paper extracts the protocol features of the data frame set by improving the AC algorithm and the association rule algorithm, using mutual information theory, and combining the definition of information correlation and redundancy with the normalization idea, further screening and verification of frame features are carried out to achieve effective extraction of protocol features, effectively improve the accuracy of feature sets, and thereby improve the efficiency and accuracy of protocol recognition.

Balanced load is one of the most important performance metrics for on-chip communication. To design flexible, adaptive load-balanced routing algorithms, an arbiter that alternated the priority order fairly is employed. With fair arbiter to decide the routing path for network-on-chip, balanced load can result in performance improvement in speed and capacity for both congestion-oblivious and congestion-aware routing schemes.

In order to solve the problem of low accuracy of small target detection in target detection, a small target detection model S-Darknet is proposed. The algorithm is designed based on the Darknet-53 network. First, a new backbone network is proposed, which fully extracts the feature of small objects and adapts to multi-scale detection. Then, in order to enhance the feature information of the target after the fusion, a feature enhancement module is added before each upsampling. Finally, the proposed algorithm was verified on the VOC2007, VOC2012 data sets and the actual data sets of railway freight locks. Experimental results show that this method has high detection accuracy under the premise of ensuring real-time performance.

Sleep Apnea Syndrome (SAS) is a serious threat to the life and health of patients. In this paper, a breathing coil is used as a flexible sensor, and a breathing signal conditioning circuit is designed to collect breathing signals. At the same time, an algorithm for determining the incidence of apnea is designed. First, all the maximum points in the respiratory signal and the corresponding extreme positions are found; then the cubic spline interpolation function is used to fit the upper and lower values of the maximum and the minimum envelope, and find the average value of the upper and lower envelopes; finally, the peak or trough point by searching for the local minimum or local maximum in the average envelope was found, and then further segmented calculations was performed. The test results show that the misjudgment range between the number of sleep apneas recorded by the simulation experiment and the number of sleep apneas obtained by the algorithm is between -2 and 3 times. The designed system terminal equipment is accurate, comfortable to wear, can be monitored for a long time. And is suitable for family daily sleep monitoring.

In recent years, face recognition technology has been widely used in security, finance, media data retrieval and other fields due to its convenience, concealment, and stability, and has become a popular research field in various scientific research institutions. Among them, face recognition under complex conditions under different lighting, different expressions, and partial occlusion has become a research hotspot. Sparse representation algorithm, as the mainstream algorithm to solve the problem of face recognition under complex conditions, although it improves the recognition accuracy compared with other algorithms, there is still room for improvement. Aiming at the partial occlusion problem in face recognition, this paper conducted many experiments under different training samples through the sparse representation algorithm of fusion error. The experimental results show that the algorithm improves the recognition rate and provides a new solution for the problem of face recognition under partial occlusion.

Oral anesthesia booster is an automatic injection system that replaces manual precise injection and solves the problem of abnormal pain caused by traditional oral anesthesia injection. Through the analysis of the system composition and model, it is determined that the pressure at the output end of the motor collected by the system in real time mainly reflects the load torque of the motor, so as to obtain the dynamic pressure detection method. According to the results of comparison with the high and low pressure thresholds, the oral anesthesia booster is controlled to select an appropriate injection speed mode. In addition, the pressure detection process is described, and the dynamic characteristics of the system are simulated and analyzed, which proves that the speed response of the system is timely after the dynamic pressure changes. The system can automatically alarm for obstruction or leakage according to the monitoring results, remind the doctor to correct the pressure and needle insertion position applied to the handle, so as to maintain the most rational needle insertion position and pressure, reduce the patient's pain, and improve the comfort of treatment.

The research is mainly based on the concept of technologies of software exploratory testing, Questing Patterns (Q-Patterns) and reusable test case, taking a reusable test case as an example, this study is devoted to the development of a feasible reusable test case management system based on exploratory testing in rapid prototype development method, which the process of design, storage, reuse and management of test cases could be implemented positively in the first system version. Furthermore, further work for technical improvement in the next version of this system is put forward.

The optimal reconfiguration of the distribution network doesn't require any cost and becomes an effective technique for reducing power loss. So this paper proposes an efficient cuckoo search algorithm for this problem. First, the mapping relationship between the cuckoo search algorithm and the distribution network is established; secondly, the algorithm generates the initial population based on the set of feasible solutions; finally, algorithm updates the nest position by the Lévy flight and nest parasitic behavior. A variable domain search algorithm is introduced to improve for the shortcomings of the cuckoo search algorithm in the process of solving the distribution network reconfiguration problem. Finally, it is verified that the algorithm has the ability to quickly solve the optimal solution by the IEEE33 model simulation analysis.

Aiming at the lightweight design of SM4 cryptographic algorithm, a lightweight cryptographic algorithm scheme is proposed and implemented in hardware. Based on the work of Li et al., a 4-bit S-box is selected. The original 8-bit look-up table S-box in the SM4 algorithm is replaced with a 4-bit cyclic shift S-box, and the same circuit structure is used to save resources in hardware implementation. A threshold scheme based on the principle of secret sharing is proposed to improve the overall security of the scheme against side-channel attacks. The input information is divided into three groups, and all operations are converted from GF(2⁴) to GF(2²) through a multiplier and a squarer, which further reduces the circuit consumption and improves the security of the algorithm. In the realization of the threshold scheme, virtual values are intro-duced to ensure that the scheme meets the uniformity, and the decomposition method is used in the inverter to re-duce the number of operations and the proportion of space. After theoretical deduction, the threshold scheme as a whole has the ability to resist first-order and second-order DPA attacks. The overall hardware implementation of the program uses a total of 2803 logic elements, which proves that it achieves the goal of lightweight while ensuring safety.

With the application and development of Internet of things technology, the demand for home intelligent devices is higher and higher. Home fire safety is particularly important. In view of the current home fire safety problems, this paper proposes a home fire alarm system based on the combination of microcontroller and sensor. The system takes microcontroller and sensor as the main components, and uses the combination of combustible gas detection module, temperature detection module and flame detection module to form a fire alarm system. The functions of gas leakage early warning exhaust, fire sound and light alarm exhaust and high temperature detection alarm are realized. Combustible gas detection, temperature detection and flame detection all collect data through sensors, and then compare with the preset alarm value to judge whether the fire occurs. The design has the advantages of accurate detection, high degree of intelligence, convenient installation, strong scalability and so on. It gives a new solution to the fire safety of the home.

The main purpose of professional English translator is to translate more ambiguity and more professional words and sentences to meet the needs of the work. This article combines web crawler based on Python language, and analyzes Python programming and online crawler implementation. This article uses this as a basis to study how to design a professional English translator, and study the structure, business workflow, and functions of the translator. The test results prove that the designed program can effectively realize the crawler function, so that it can obtain a large amount of consistent data in a very short time, and return these related data to the user through the interface connection design. This can not only meet the needs of users for quick translation and search of English words and sentences, but also meet the needs of actual work.

This article mainly briefly introduces the related content of computer-aided innovation technology. This article discusses the effective application of computer-aided innovation product design technology by analyzing the necessity of knowledge utilization in the process of product innovation design. This article takes full advantage of computer technology, constantly innovates product design, develops products that meet people's actual needs based on market demand, and optimizes product design. This can guarantee the quality of product design and enhance the company's market competitiveness. The effective use of key knowledge can not only solve the technical difficulties in product design innovation, but also improve the technical level of computer-aided innovation product design.

Aiming at the global path planning problem of unmanned underwater vehicle (UUV) in three-dimensional environment, a fusion algorithm named PACO based on particle swarm optimization (PSO) and ant colony optimization (ACO) is proposed. In order to meet the needs of three-dimensional path planning, the grid method is used to model the actual marine environmental data. Besides, PSO and ACO are optimized respectively. Taking the path length as the measurement index, the simulation is realized by using MATLAB. The simulation results show that the initial path search ability of the fusion algorithm is significantly improved, and the effectiveness of the algorithm is verified.

Firstly, Logistic model and parameter estimation are simply introduced. Secondly, we make some analysis and summary of various methods of solving Logistic model estimation, which have been put forward by scholars in recent years. These methods include: using Logistic model's integral form, linearization, optimization method, numerical method, grey system dynamic model, population model to solve the modeling parameter identification method. What's more, the practical applications of Logistic model are introduced such as the application in the process of urbanization.

For the research of Tibetan scene text detection and recognition, it is time-consuming and laborious to collect and annotate natural scene data labels manually. Therefore, artificial synthetic data is of great significance for promoting relevant work. This paper focuses on the study of replacing other languages in the scene with Tibetan, while maintaining the style of the original text. We decompose the problem into three sub-networks: text style transfer network, background inpainting network and fusion network. Firstly, the text in the foreground image is replaced by the text style transfer network to generate the foreground image. Then the background inpainting network erases the original text in the style image, and uses the surrounding information around the text to fill the text area to generate the background image. Finally, the generated foreground image and background image are used to generate the target image by the fusion network. We experimented with conversions from English to Tibetan and English to English to verify the generalization and robustness of the network. Experimental results show that its accuracy (SSIM, PSNR) on some datasets (SVT, ICDAR 2013) has been improved to some extent.

Pedestrian detection is mostly used in autonomous driving scenarios, which require high real-time detection. Most of the existing algorithms are based on Two-Stage detection, with poor real-time performance. This paper proposes a pedestrian detection system based on One-Stage, and uses One-Stage-based YOLO-Tiny, YOLO and YOLO-SPP algorithms to test and analyze the system in the scene of detecting pedestrians of different sizes. The results show that YOLO and YOLO-SPP have high average confidence, and YOLO-Tiny has a fast detection speed, which is suitable for fast detection scenarios.

When carrying out English translation based on phrases and sentence patterns, the contextual coherence of phrases and sentence patterns is relatively poor, which will directly affect the effect of English translation. In the process of designing and improving the English translation computer intelligent proofreading system, we need to study the specific modules of the system design. At the same time, we need to improve the design level of the English translation computer intelligent proofreading system based on the improved phrase translation model. Only in this way can it be ensured that vocabulary can be proofread when translating English, and the translation effect can be improved. This is also the main goal to realize the design of an intelligent proofreading system for English translation. In the process of analyzing and researching the proofreading system, we need to analyze and compare the accuracy of the design system and proofread the incoherent context in the English translation results in time. This can give full play to the advantages of the intelligent angle system with high accuracy and contextual coherence.

With the arrival of the era of big data, the data generated by the monitoring system grows rapidly, the storage and quick retrieval of massive monitoring data pose a challenge to the database. HBase is an open-source version of Bigtable, built on top of HDFS, based design and query on rowkey is quick, but with the development of database application, when retrieving non-rowkey fields, HBase can only query by scanning the full table, which obviously cannot meet the actual requirements, and the query of the monitoring data is mostly based on the most recent data, so for this problem, this article presents a secondary indexing approach based on coprocessor and Redis. This method realizes the rapid creation and automatic update of secondary indexes in HBase and Redis through the coprocessor. It gets the corresponding rowkey quickly during retrieval, and query the corresponding data in the data table according to the rowkey. The former stores the index of the data table, while the latter stores the index of the latest data in the data table, to improve the real-time performance of data retrieval.

Given the current complex heterogeneous network environment, choosing different networks in different user scenarios limits the effectiveness of network access. To solve this problem, an algorithm based on DDQN is proposed to optimize the network configuration. The reward value is determined by service type, throughput, delay, and signal strength. DDQN can avoid the ping-pong effect that can be generated by network selection and stabilize the network environment by splitting action selection and action evaluation to prevent overestimation. The experimental results show that the algorithm can autonomously select and assign network protocols and can more accurately find network access methods suitable for user scenarios than the current network switching method represented by a single network parameter based on RSSI and CCA.

There are many power equipments in the UHV valve hall, including the converter transformer, converter valve tower, all kinds of bushing, tubular bus. The operation condition monitoring of the above main equipment mainly includes electrical performance parameter measurement, oil and gas insulation medium performance parameter measurement and other means. This paper mainly introduces a monitoring method for the operation status of converter valve hall from the perspective of image processing technology. The technology mainly includes: 1) the application of intelligent image processing technology to identify and classify the database of infrared thermal imager and ultraviolet imager; 2) the real-time on-line measurement for the insulation distance of typical fittings through Kalman filter technology, including the measurement of the hottest temperature The insulation distance between partial discharge point and zero potential grounding point; 3) the electric field simulation model of typical main equipment in valve hall is established based on three-dimensional modeling technology of finite element method, and the surface electric field distribution of key main equipment fittings is obtained. Combined with the image database information, insulation distance information and typical main equipment electric field distribution information, the operation state parameters are effectively obtained, and the intelligent algorithm is applied to automatically evaluate its operation state, discover the latent fault and locate the positive fault, so as to provide effective data support and the protection strategy for the main equipment operation and maintenance.

The data processing and analysis flow of Internet of things system is the basic process of system design ideas and methods. It includes a series of operations such as requirement analysis, software function design, algorithm and method of implementation, overall structure design, module design, coding and debugging, program linking and testing, compilation and submission of the system. Data processing and analysis process can meet the needs of the system, and solve the problem of the main content of the process. This paper discusses the basic mechanism, principle, type and application of data processing of Internet of things, discusses the composition and elements of data analysis system of Internet of things, and explains the construction process and analysis process and steps of data processing of Internet of things.

In order to study the relationship between rainfall and related meteorological elements in rainy weather, find out the changes of related meteorological elements before and after rainfall, a multi-dimensional time series data mining model is proposed. The model first performs dimension selection pre-processing on the time series of meteorological elements to remove irrelevant and redundant dimensions, then uses the proposed extreme slope piecewise linear fitting method to segment the time series, data compression and eigenvalue extraction, and finally uses k-means clustering algorithm to symbolize the processed multi-dimensional sequence, and uses rules to extract the rainfall weather model. Experimental results show that the model has good practical value.

With the continuous development of geographic information system technology, the ability to analyze and process various spatial information has been greatly enriched and strengthened. The optimal path analysis is the current research hotspot in spatial analysis. In this paper, Changsha City in Hunan Province was taken as an example to build an optimal path selection platform based on WebGIS. With the help of ArcGIS network analysis tool, network analysis data set was established for vector road data, and road length was selected as impedance factor to realize optimal path search.

Combining with the actual work of the poverty aid management in F University, this paper analyzes the work flow of the poverty aid management in University, designs and develops the Poor University Students Management System. The system uses web architecture mode, JSP programming technology and SQL Server database management technology. The system is composed of four modules: system management module, institute management module, class management module and student management module. It realizes the functions of poor students' information collection, poor students' application, students' information query and poverty level audit. It improves the efficiency of poor students' management and realizes the networking, automation and informatization of poor students' management.

Load balancing technique plays a fundamental role in distribution, while facing numerous challenges. It is difficult for ordinary load algorithms to cope with complex scenarios, and it is likely that uneven load will lead to downtime of a single server, thus causing an avalanche effect and eventually breaking down the whole cluster. This paper proposes a load balancing strategy based on improved ant colony algorithm, considers the influence of task length and processing efficiency in the algorithm based on the original ant colony algorithm, and takes the load rate of nodes as an important index for ants to select path. In addition, the load coefficient includes CPU, memory, IO, and network bandwidth into the calculation range, which makes the obtained server load condition more accurate. Finally, test data of different algorithms are obtained through experiments, and the advantages and disadvantages of this load balancing strategy are verified after horizontal comparison.

In order to enable the microgrid to meet the system load demand while performing economically optimal operation scheduling, this paper establishes an island-type microgrid model, which is optimized by using an improved immune particle swarm algorithm, and the inertia weight and learning the two parameters of the factor are improved. On the basis of the immune particle swarm algorithm, a power exponential function operator is added to the inertia weight to improve the search ability of the algorithm, in order to reduce the computing time, the dynamically adjusted learning factor is introduced to optimize the immune particle swarm algorithm the local search ability is stronger. Two examples are selected to verify the algorithm, the results prove that the method has better global convergence and local search capabilities and the convergence speed has been improved.

The rapid development of science and technology makes the use of new media technology more and more common. In the 21st century, the communication mode of new media is becoming more and more diversified, and its convenient and efficient communication mode can meet the needs of people's modern office and life. Digital audio has a variety of production technology, and the production accuracy and controllability are very high, so it occupies the main position in the information composition of new media. As a sensory media, sound is mainly manifested in voice and music in daily life. At the same time, computer music production technology mainly includes digital processing technology and MIDI production technology. It can be seen that computer music production technology has become the core technology of sound editing, production and processing under the background of new media. This paper takes the application of computer music production technology as the research object, analyzes the internal relationship between new media and music production technology, the application and development of music production technology and so on, so as to provide the development direction for computer music production technology.

Extreme operational conditions, e.g., wind storms and ice storms, have been posing great challenges for power grid operation. The extreme conditions are likely to cause widespread blackout, therefore a black-startprocedure is needed to restore the non-blackstart generating untis and transmission assets. To falicilate the balck-start procedure, the coordination between post-event preventive dispatch and post-event black-start dispatch is established, considering the uncertainty of forced transmission line outage scenarios. A two-stage robust optimization model is built to coordinate the preventive dispatch and black-start procedure. An iterative decompostion algorithm is applied to solve the proposed model. A synthesized Chinese 372-bus system is used to illustrate the proposed method. It is shown that, in the worst-case blackout scenario, the coordinated black-start procedure can accelerate the entire process by 15% and reduce the black-start capacity requirements.

As an important part of energy Internet construction, power consumption information acquisition system, as the "total entrance of power information data acquisition and the total exit of control instruction execution", plays an increasingly important role in the safe and stable operation of the system. At present, the on-site access equipment is diversified, but there is a lack of unified active security defense early warning module, and there is no perfect security monitoring system. There are security risks such as illegal replacement of on-site terminals and private upgrade of terminals. In order to further improve the security and stability of smart grid operation, this paper makes an in-depth study on the existing security problems of the acquisition terminal and the online security monitoring of the acquisition terminal, and adopts the device fingerprint technology and password technology to meet the needs of the security protection of the acquisition terminal.

The establishment of a basic service platform for the Internet of Things application can provide the necessary basic technical services for the application of the Internet of Things in various industries in the future. In this way, the in-depth integration of data and business levels in various industries can be realized, so as to fully realize the interconnection of IoT applications. This article will analyze and introduce the technology, principles, and related platform design and deployment required to build a practical platform for IoT application software on the basis of an overview of the technical characteristics, content and composition of the Internet of Things.

The current field of lipreading is limited to the processing of visual signal and the optimization of sequence models, but the sentence text is ignored. Aiming at this problem, we proposed a lipreading method combined with natural language processing (NLP) technology, Lip-Corrector, which applies the BERT model in this paper. The front end of the model uses 3D+2D convolutional neural network (CNN) to extract lip information, the middle end uses the Transformer-based Seq2seq sequence model to make sentence-level predictions, and the back end uses a sentence correction method based on the BERT model, which connects to the midend after pre-training on the self-made dataset. Experiments on the two largest sentence-level lipreading datasets of LRS2 and LRS3 show that the performance of this model surpasses all the baselines, which proves that lipreading methods combined with NLP technology will get better results.

With the safe and stable operation of the new generation of power line carrier communication, it is of great significance to the rapid development of smart power grid. However, due to its own physical characteristics and the impact of the external environment, communication cables usually have some security risks, such as aging, damage, theft. In this paper, a new generation of power line carrier communication trusted access network is studied. In this paper, the current situation of the new generation of power line carrier trusted access and network requirements are analyzed, then the principle and characteristics of basic ant colony algorithm are briefly introduced, then the characteristics of power line carrier communication system network and network requirements are analyzed, the improved ant colony algorithm is applied to network optimization of the new generation of PLC communication system. The algorithm makes the power line carrier communication system have certain destruction resistance and recombination ability when the network state changes, it can guarantee the reliable operation of power network communication system.

The video surveillance is a key technology, which is one of the design technology for the ubiquitous perception and acquisition of panoramic information in the smart grid. The remote video surveillance technology and power automation systems such as remote control, telemetry, remote sensing and remote signal, which ensure the normal operation of the whole smart grid, can greatly improve the security of important substations or unmanned substations. At present, there is not a unified standard for video surveillance layout of the smart grid. The video surveillance system in each area is often arranged with reference to the requirements and habits of the operation and maintenance in the region. But, other important factors are not taken into account. In this paper, by constructing the objective function of multi-dimensional evaluation, the optimal layout scheme of video surveillance in the smart grid is obtained. Finally, the effectiveness of the proposed method is proved by the analysis of the cases.

We propose a simple dense convolutional neural network model (SDenseNet) to solve the problem of low accuracy and poor performance in classification of navel orange with large external differences. Specifically, AlexNet is adopted as the backbone network, through the introduction of batch normalization(BN) and reduction initial size of the convolution kernel to accelerate convergence of the network model. Besides, we design a new feature reuse structure to promote the connection between layers, and modify the full connection layer by using global pooling to reduce training parameters. The experimental results demonstrate that our proposed SDenseNet significantly improves the performance with 99.33% accuracy on self-built navel orange dataset, outperforming the classic models of LeNet, AlexNet, SqueezeNet and ResNet with significant improvements of 5.33%, 1.55% and 3.55% respectively. The research results provide a new solution for external quality automatic detection and classification of navel orange.

Dialogue State Tracking (DST) is the core component of the task-oriented dialogue systems. Although recent neural DST models have made great progress, they often ignore the phenomenon that the current dialogue state is closely related to the earlier dialogue states. In this paper, we try to introduce the slot embedding into the transformer to focus on those special tokens, which ever appear in earlier dialogue states. On the basis, we leverage the copy mechanism to predict the state over the dialogue utterances. Our model also imitates the architecture of reading comprehension model to make full use of the current utterances. The experimental results verify the benefits of the slot embedding, and our model achieves significant improvements than baselines on MultiWOZ 2.0 and MultiWOZ 2.1 datasets.

Aiming at the problem of ORB feature matching algorithm extracting background pixels as feature points and matching wrong feature points in a complex background environment, an improved ORB algorithm based on adaptive threshold is proposed, and GMS algorithm is used to screen out mismatches in the feature matching stage. First, the algorithm calculates the mean and standard deviation of the image to be matched and the reference image. Then it inputs the obtained data into the adaptive threshold calculation stage to obtain the adaptive threshold. Finally it inputs the adaptive threshold into the feature extraction and matching stage. The experimental results show that the improved ORB algorithm reduces the number of features extracted from the background of the exhibits in the complex environment of the museum, and the matching algorithm combined with the ORB algorithm and GMS increases the correct matching on the basis of slightly shorter time than the original algorithm. The algorithm has strong robustness and real-time performance.

Multi-object tracking in autonomous driving aims to represent trajectories of moving objects for planning system of the vehicle. In this paper, we propose a new tracking-bydetection scheme based on deep neural networks for multi-object detection and tracking in autonomous driving scene. We first introduce a light-weight neural network branch for fast object detection, and based on the detection results on each frame, we build two object graphs for consecutive frames separately, where the vertices of the graph represent the objects in the image, and the edges of the graph represent the spatial relations between the objects. We then formulate the multi-object tracking problem as the graph matching process by learning the relevance between objects from another object association network branch. Experiments results on the MOT multi-object tracking dataset show that the proposed object detection and tracking approach achieves comparable results with state-of-the-art deep learning based multi-object tracking methods, and outperforms them in tracking efficiency, which ensures real-time multiobject tracking for autonomous driving.

Because of uneven distribution and indistinct boundary in imbalanced dataset, imbalanced dataset classification algorithm based on neighbors density support vector machine (NDSVM)is proposed. In this algorithm, the neighbor range density of each sample in the majority class is calculated firstly. According to the density value, the data which on the majority class border or close to the border is equal to the minority samples in quantity, which are selected, then the minority class complete SVM initial classification. Then the resulting support vector machine and residual data in the majority class optimize the initial classifier. The simulation results of experiments on the manual and UCI dataset show that compared with WSVM、 ALSMOTE-SVM and SVM, NDSVM has better classification performance, which effectively improve the classification performance of SVM algorithm on the uneven distribution and indistinct boundary in imbalanced dataset.

An improved firefly algorithm for solving the economic dispatch problem of power system is proposed. The algorithm considers the climbing rate, the dead zone and other constraints, and calculates the network loss. This algorithm is based on the firefly algorithm, a new repair strategy for active correction of the violation of the constraint condition of all kinds of fireflies, and combined with the penalty function technique, the fireflies in the feasible region or as close as possible to the feasible solution search in the area. The accuracy and speed of the algorithm can be effectively improved by greatly reducing the probability of finding the optimal location of the firefly in the infeasible solution region. The simulation results show that the algorithm has the characteristics of high speed, high accuracy and good convergence.

When the traditional methods are used to analyze the dynamics of the robot arm structure, the stability of the arm structure is not good and the simulation effect of the operation process is not good. For this reason, Kane method is introduced in this paper to carry out dynamic simulation analysis of industrial robot arm structure. The operating process of the industrial robot arm structure was simulated by Kane method, and the angular acceleration of the robot's main arm, forearm and push rod was solved in each cycle. The Jacobian matrix of the 5 degree of freedom manipulator joint is established, the singular configuration of each joint is calculated, and multiple joint variable combinations are generated by random number. The forward motion equation of the robot endpoint set was solved to obtain the endpoint set of the robot. The Lagrangian method was used to model the robot with Kane 3D simulation software to simulate and analyze the dynamic process of the driving torque required to grab the mass workpiece at the three joints. The joint motion model of robot is established on ADAMS simulation software, and the torque curve of the rotating joint of industrial robot arm structure is obtained. MATLAB simulation software is used for simulation. The results show that the integrated design method of manipulator and machine tool is correct, which provides a comprehensive data basis for the selection of servo motor.