A method for testing partial discharge characteristics of switch cabinet based on improved particle swarm algorithm

In the current operational environment of the switch cabinet, the presence of mixed-frequency signals poses interference from white noise to partial discharge detection, resulting in suboptimal detection efficiency and accuracy. Therefore, this study explores a method for testing the partial discharge characteristics of switch cabinets based on an improved particle swarm algorithm. This novel approach involves signal processing of partial discharge in a switch cabinet, feature extraction based on the particle swarm algorithm, and feature output for partial discharge detection. Experimental comparisons demonstrate that the new detection method significantly enhances detection efficiency and ensures the accuracy of results, meeting the operational requirements of substations.


Introduction
Switch cabinets hold a paramount position within the realm of electrical power systems, serving as the linchpin in the intricate network of power generation, transmission, and distribution.Their significance lies in their multifaceted role, which profoundly impacts the reliability, safety, and efficiency of grid operations [1].
One of the fundamental functions of high-voltage switchgear is current control.It ensures that electricity flows smoothly through the network, preventing overloads and voltage fluctuations [2,3].This aspect is crucial in maintaining the quality of power supplied to end-users, including residential, commercial, and industrial sectors.Current control is essential for preventing disruptions and equipment damage.
Fault protection is another critical role of switch cabinets.In the event of electrical faults, such as short circuits or equipment failures, switchgear swiftly isolates the problematic segment, thus mitigating the risk of widespread power outages [4].This swift response to faults enhances grid reliability and minimizes downtime, ultimately ensuring a continuous supply of electricity to consumers.Equipment isolation is an indispensable feature of high-voltage switchgear.It allows for the safe maintenance and repair of electrical components [5].By providing isolation points, switch cabinets create a secure environment for maintenance personnel to work on various equipment without exposing them to live electrical circuits.This not only enhances the safety of maintenance procedures but also ensures the efficient and manageable operation of the power system.
In the context of power asset monitoring, high-voltage switchgear plays a pivotal role in the oversight and management of electrical assets [6].Advanced protection relays and monitoring systems are integrated into switch cabinets, enabling real-time monitoring of key parameters.These systems swiftly detect deviations from normal operating conditions and respond by taking preventive actions.The result is a reduction in equipment damage and an extension of the operational lifespan of critical assets, all of which contribute to grid efficiency and reliability.
As technology continues to advance, the capabilities and features of high-voltage switchgear are evolving to meet the growing demands of the power industry.Modern switch cabinets are becoming increasingly intelligent, incorporating digital control systems and communication interfaces that facilitate remote monitoring and control [7,8].This evolution enables more precise and efficient grid management, accommodating the integration of renewable energy sources and emerging smart grid technologies.Nevertheless, the efficacy of these sophisticated detection technologies can be compromised by external mixed-frequency signals and white noise, leading to reduced detection precision and efficiency [9].To address this challenge, researchers have turned to the particle swarm algorithm.This innovative method excels in recognizing and analyzing characteristic parameters, constructing fuzzy information analysis models by identifying various key features, and thereby obtaining more stable signal characteristics.By harnessing the capabilities of the particle swarm algorithm, research efforts are aimed at improving the accuracy and efficiency of partial discharge detection methods for switch cabinets.
In conclusion, high-voltage switchgear is a cornerstone of modern electrical power systems, integral to the safe, reliable, and efficient distribution of electrical energy.As the power industry continues to advance and evolve, switch cabinets are at the forefront of innovation, offering enhanced features and capabilities.The ongoing research into detection methods, such as the particle swarm algorithm, underscores the commitment to further improving the performance and reliability of these critical components in our power infrastructure.

A method for detecting partial discharge characteristics of switch cabinet based on particle swarm algorithm
To meet the requirements for detecting partial discharge in electrical equipment, it is essential to conduct a parameter analysis of the coupling process for partial discharge in electrical equipment, taking into account load fluctuations, prior to designing the method.Based on the constraints imposed by operational parameters during the electrical equipment's operation, a parameter identification model is constructed [2].Through this approach, an output model for the maximum power parameter during equipment operation can be obtained.Under these conditions, the parameter constraints can be represented by the following computational formula: where A represents the constraints on operational parameters for the switch cabinet, e represents the control parameter, typically assumed to have a constant value, i denotes the harmonics generated during the transmission of partial discharge signals in the operation of the switch cabinet,  signifies the electrical signal capacity of the equipment, and t indicates the frequency of locally generated electrical signals [9].
where G represents the electrical signal characteristics, F represents the disturbance compensation of the electrical signal, and S signifies the amplitude fluctuation of the signal, with S constrained within the range of +1 to -1.After signal extraction is accomplished, further processing is undertaken.During this processing stage, a fuzzy parameter recognition method is introduced to constrain the extracted feature information.This results in the derivation of localized discharge characteristic information.The controlled discharge is represented by the following computational formula [3]: where H denotes the controlled discharge signal, t represents parameter reliability, and  represents the fuzzy characteristic quantity of the signal.Parameters with similar characteristics are fused to construct a completely new electrical signal, which is then output to finalize the signal processing.

Improvement of the basic particle swarm algorithm
The Particle Swarm Optimization (PSO) algorithm is a classical method for solving optimization problems.It starts from random solutions and iterates multiple times to converge to the optimal solution.The specific description is as follows: In a D p -dimensional space, a population is formed by consisting of n p particles [10].In the d-th dimension of this space, the position and velocity of the i-th particle are represented as x id and v id , respectively.Initially, the fitness value corresponding to each particle's position x id is calculated, and by comparing these values, the current individual best solution P id is obtained.Subsequently, starting from the position of the best solution, the algorithm seeks the global best solution P gd .During the iteration process, particles update their positions x id and velocities v id to search for the optimal solution, and both P id and P gd are continuously updated.The formulas for updating particle positions x id and velocities v id are as follows [4]: The precision of the solutions obtained by the basic particle swarm algorithm is not necessarily directly proportional to the number of iterations and the size of the particle swarm.In other words, increasing the number of iterations and the size of the particle swarm does not necessarily lead to higher solution accuracy [5].This is directly related to the initialization of particles as random solutions, which can significantly impact solution accuracy and iteration speed.To address this issue, this paper proposes an approach where, during the initialization of particles, high-quality initial particles are first obtained through the method of evenly distributing virtual inertia within a wind farm.Subsequently, the algorithm iteratively seeks the optimal particle based on this initial set, which can improve the algorithm's solution effectiveness [6].
Algorithm-solving steps are outlined as follows: Step 1: We compute high-quality initial particle positions and proceed to initialize the particle swarm, including parameters such as population size, maximum iteration count, particle positions, and particle velocities.
Step 2: We calculate particle fitness values (maximum system frequency deviation) by using the initialized data, and compare and replace these values with the individual best values obtained.Next, we compare and replace them with the global best values within the particle swarm.Finally, we update particle positions and velocities based on Formulas (4) and (5).
Step 3: We determine if the termination conditions are met.If the minimum value meets the accuracy requirements or the maximum iteration count is reached, we output the optimal value.If the conditions are not met, we return to Step 2 to continue the iterative computation [7].

Output of local discharge detection features
Upon completing the aforementioned design, an information analysis model tailored for the detection of partial discharges in the switch cabinet is established.The discharge layer is denoted as y, with corresponding values ranging from 1 to 0. The output of local discharge characteristic functions, represented as Z, can be described by using the following computational formula [10]: where Z represents the detection feature output function, and y represents the output quantity.Based on the output results of different detection axis frequency components, localized components are subjected to detection, and detection results are generated.Subsequently, the output results undergo coupling processing, which is outlined in Formula ( 7).
(1) Computational formula: As mentioned earlier, there is a computational formula that relates the discharge layer y to the output of local discharge characteristic functions Z.This formula likely involves various mathematical operations and possibly combines data from different detection axes.
(2) Coupled outcomes: The results obtained from different axis detection are processed and combined to produce "coupled outcomes".This implies that information or data collected from multiple detection axes are integrated to create a more comprehensive view of the discharge signals.This integration may involve statistical analysis or other techniques to extract relevant information.
(3) Parameter components: These coupled outcomes generate a set of "parameter components".These components likely represent specific features or characteristics of the discharge signals.These parameters serve as descriptors of the discharge signals and can capture important information about their nature, magnitude, or other relevant attributes.
(4) Adjustment for detection: The parameter components can be adjusted or manipulated to finetune the detection of discharge characteristics in the switch cabinet.This adjustment process may involve setting thresholds, tuning algorithms, or employing machine learning techniques to identify and classify discharge events based on the extracted parameters [11].To further substantiate the efficacy and practical applicability of the detection method proposed in this study, signal data collected from the operation and maintenance of a specific 35 kV substation was utilized as the experimental dataset, as illustrated in Figure 1 [12].After the application of both the detection method introduced in this paper and a Geographic Information System (GIS)-based detection method, detection maps were obtained.The main focus of this comparative analysis was on discerning the local discharge characteristics of the switch cabinet.Two critical evaluation criteria, namely detection latency and localization error in post-detection, were adopted for the assessment.As outlined earlier, experiments were carried out on five distinct switchgear, and the results of these experiments have been compiled in Table 1 for analysis (also presented in Figure 2 and Figure 3).The table presents a comparative view of the experimental outcomes between the detection method relying on the improved particle swarm algorithm and the GIS-based detection approach.The metrics of detection delay (expressed in seconds) and localization error (in centimetres) are presented for each of the five switchgear.  1 underscores a significant advantage of the detection method proposed in this study.Specifically, when compared to the GIS-based detection method, the proposed method based on the improved particle swarm algorithm demonstrates substantially smaller detection latency and lower localization error values.These findings indicate a marked improvement in detection efficiency and accuracy.

Comparative experiments
Moreover, the experimental results suggest that the proposed method is minimally affected by external interference factors during practical applications.This characteristic enhances its reliability and ensures consistent detection efficiency and precision.The lower latency and reduced localization error values further affirm its suitability for real-world applications, particularly in the context of switch cabinet condition monitoring and maintenance.This research highlights the promise of the improved particle swarm algorithm as a valuable tool for enhancing the accuracy and efficiency of partial discharge detection in switch cabinets.

Conclusion
This paper has designed a novel detection method based on the particle swarm algorithm.After completing the design, comparative experiments have demonstrated that the proposed method performs relatively better in practical applications.It can achieve precise detection of electrical signals under conditions that largely eliminate external interference.This method can be utilized for remote-assisted detection during on-site operations and assist field technicians in decision-making.However, for widespread adoption of this method in the market, further testing and refinement of its performance are needed in subsequent designs.It is hoped that through this design, the optimization of the method's effectiveness in subsequent use can be achieved.

Figure 1 .
Figure 1.Switch cabinet partial discharge test platform.To further substantiate the efficacy and practical applicability of the detection method proposed in this study, signal data collected from the operation and maintenance of a specific 35 kV substation was utilized as the experimental dataset, as illustrated in Figure1[12].After the application of both the detection method introduced in this paper and a Geographic Information System (GIS)-based detection method, detection maps were obtained.The main focus of this comparative analysis was on discerning the local discharge characteristics of the switch cabinet.Two critical evaluation criteria, namely detection latency and localization error in post-detection, were adopted for the assessment.

Table 1 .
Comparative Experimental Results of Two Detection Methods.