Research on Application Scenarios of Artificial Intelligence in New Power System

Building a New Power System with new energy as the main body is the main theme of modern energy system construction and energy for a period of time in the future. It is an inevitable choice to promote the clean and low-carbon development of electric power. Through artificial intelligence technology, we can solve the problems of strong volatility and difficult regulation caused by the diversity of new energy sources, and promote the establishment of a green, efficient, flexible, interactive, safe and controllable New Power System. This paper sorts out the challenges and deficiencies faced by New Power System, summarizes the application characteristics of artificial intelligence in various industries, analyzes the development and characteristics of artificial intelligence applications, combines the application hotspots of artificial intelligence New Power System, studies the application scenarios of artificial intelligence in the construction, operation and maintenance of New Power System, and puts forward suggestions on the research direction.


Introduction
With the proposal of the goal of "Carbon peaking and carbon neutrality" in China, smart energy has become one of China's development strategies.At the ninth meeting of the Central Committee on Finance and Economics, it was proposed that the construction of a new power system is an inevitable way to ensure the rapid and high-quality development of new energy, as well as a long-term and systematic project.In October 2022, the 20th National Congress of the Communist Party of China proposed that we should build a strong digital country, deepen the energy revolution, speed up the planning and construction of a new energy system, and put forward higher requirements for promoting the digital transformation of energy.With the exploration of clean energy, the construction of distributed energy and the development of new energy equipment, the energy management and control system with the characteristics of new energy production, storage and consumption is becoming more and more important.The modern power grid system presents various unstable characteristics, such as diversification, complexity and high difficulty in balancing.The requirements are getting higher and higher for stable operation of the power grid system.Artificial intelligence is an effective way to solve multi-point balance, large data computing and complex system control [1] .Combining artificial intelligence technology with stable operation, flexible dispatch and automatic maintenance of emerging power systems, can not only ensure the safe and reliable operation of the system, but also measure and analyze the important operation data in the system in real time, which can form a sustainable development path of high efficiency and energy saving.

Necessity of New Power System development
Speeding up the construction of a new power system with new energy as the main body has important guiding significance for the transformation and development of energy and power in China.The 14th Five-Year Plan for Modern Energy System proposed to build a new power system and promote the evolution of the power system to adapt to large-scale and high-proportion new energy.Building a new power system with new energy as the main source is the core task of the energy and power industry to serve the "Double carbon" targets.To achieve the "Double carbon" targets, energy is the main battlefield, and vigorously developing new energy is the key.According to the statistics of the National Energy Administration, in 2022, the newly installed capacity of wind power and photovoltaic power generation in China reached 125 million kilowatts, and the newly installed capacity of renewable energy in the whole year reached 152 million kilowatts, accounting for 76.2% of the newly installed power generation in China, which had become the main body of newly installed power in China.Renewable energy generation reached 2.7 trillion kWh, accounting for 31.6% of the total electricity consumption of the whole society, with increasing of 1.7 percentage points over 2021.The role of renewable energy in ensuring energy supply is becoming more and more obvious.
With the new energy gradually becoming the main power supply, power system security and stability control will face unprecedented challenges.It is an urgent need to build a more powerful, more flexible and more resilient new power system.The deep integration of low-carbon energy technology, advanced transmission technology and advanced information and communication technology, network technology and control technology not only strongly promotes the rapid development of new energy, distributed generation and microgrid, but also greatly promotes the improvement of system analysis and prediction ability and operation control level [2] .

Connotation and characteristics of New Power System
Xin Bao'an, chairman of the State Grid, believes that it is necessary to break through the construction of new power system from five aspects: innovating the development of power grid, enhancing the regulation ability of power grid system, overcoming technical difficulties, diversifying the construction of power grid and uniting the development of industry.Based on the views of experts in the industry, the new power system is a new generation of clean and low-carbon, safe and controllable, flexible and efficient, intelligent and friendly, open and interactive power system with new energy as the main body and digitalization as the means.
The new power system has five characteristics: clean and low-carbon, safe and controllable, flexible and efficient, intelligent and friendly, open and interactive.The new power system has objective, measurable and controllable characteristics, toughness, flexibility and strong self-healing ability, strong transmission and distribution side regulation ability and resource allocation ability, flexible and efficient conversion of various energy sources.The biggest difference from the traditional power system is that it is highly digitalized and networked, friendly and coordinated with the source, network, load and storage terminals, adapted to the large-scale access of various high-tech and equipment, closely integrated with the power market, actively responded and interacted with each other [3] .

Challenges to New Power System
China's electricity demand will maintain a linear growth trend in the next few decades.According to the forecast of the Power Supply and Demand Research Laboratory, China's social electricity consumption is expected to meet 13.6 trillion kWh by 2050, requiring the energy supply and marketing of the power system to maintain long-term stability, sustained safety, clean and efficient, which makes the construction and operation of the new power system face many challenges.
• For highly fluctuating and intermittent wind power and photovoltaic energy, there is a lack of flexible power regulation capabilities.• The strong randomness of new energy and the difficulty in regulation of source, network, load and storage bring challenges to the real-time balance regulation of the power system.
• The equipment and operation safety of the system are facing challenges.
• The increase in the interaction between supply and demand on the power consumption side brings new challenges to the adjustment capability and service model of the power system.
• The lack of multi-dimensional digital empowerment in the whole process cannot support the highly intelligent operation of the power grid.The new power system using artificial intelligence technology can allocate energy flexibly and stably, coordinate the coupling effect of new energy access, ensure the safe operation of the system equipment, and reduce the peak load and construction cost of power system.

Overview of Artificial Intelligence
Artificial Intelligence is a development route of computer intelligent application.At present, the mature AI technology mainly includes speech recognition, machine learning, biometric recognition and so on.According to the research and analysis of Stanford DAWNBench team, the cost of developing AI projects today is only 1/150 of that in 2017.As the cost has fallen, the scale of investment has risen rapidly.According to the statistics from the White Paper on Artificial Intelligence (2022) by the China Institute of Information and Communications, in April 2021, the European Union invested about 7.59 billion euros in digital technology, including AI; in 2021, the United States invested about 1.5 billion dollars in non-defense AI, and the United Kingdom invested more than 2.3 billion pounds in AI from 2014 to 2021 [4][5] .

Characteristics and advantages of Artificial Intelligence
The characteristics of AI: First, Artificial Intelligence perceives the external environment through receiving components, and collects, analyses and processes data information through text, voice and action, to produce valuable information flow and knowledge model.Second, the autonomous learning ability of artificial intelligence can realize the iterative evolution of machines and systems in the digital age and adapt to the changing external environment.Third, Artificial Intelligence integrates development process and model management efficiently, and establishes a research and development system of AI engineering tools, including autonomous machine learning, external perception system, precision mechanical control, human-computer interaction, etc.
The advantages of artificial intelligence: First, with the continuous updating of super-large-scale algorithm training, the basic computing power of artificial intelligence is developing towards customization and diversification, and data processing, storage and interactive computing technology to promote the upgrading of computing technology, resulting in refined and customized data services.Second, with the increase and upgrade of sensors and intelligent terminals, human-computer interaction forms are becoming more and more abundant.Artificial Intelligence has more ability to judge, thus providing more humane services.Third, artificial intelligence makes all industries pay more attention to service experience.Integrating product services with the help of artificial intelligence has become an inevitable trend of industrial development and product value-added.

Application of Artificial Intelligence in Different Industry
With the maturity of artificial intelligence technology, the improvement of algorithms, the reduction of costs and the acquisition and application of massive data, technologies such as machine learning, knowledge mapping, natural language processing, human-computer interaction, computer vision, biometric recognition are being rapidly applied to government, Internet, finance, energy, medicine and other industries.More and more repetitive work and fixed process work are gradually replaced by artificial intelligence.See Table 1  Using historical data from the data center to train the neural network system, predict the temperature and workload in the next few hours, and push out energy-saving schemes in real time to improve the accuracy of energy consumption prediction, DeepMind has used this technology to cut the cooling energy cost of Google's data center by 40%.

5
Weather and atmospheric forecasting techniques Nnergix (Spain) Machine learning techniques are used to predict the impact of atmospheric and weather conditions on renewable energy capacity, such as the hourly output of photovoltaic power plants.
In the field of power grid, there are many applications in transmission, substation, distribution, infrastructure, operation and maintenance, marketing management, power trading and other scenarios.However, the large-scale application of artificial intelligence in the field of power grid still faces technical bottlenecks.The applications are mainly distributed in their relatively independent professional or technical routes, which can not form the whole intelligent application environment of the whole chain of power grid system, and can not play the greatest role of artificial intelligence in the direction of universalization and energy Internet.

Analysis based on Complex Adaptive System Theory
Professor John Holland of the United States formally put forward the theory of complex adaptive system in his book《Hidden Order: Adaptation Creates Complexity》.The theoretical definition is that there are mutual influence, interaction, feedback and other associations between multiple subjects or between multiple subjects and the environment, and a new environment is formed through the association of specific rules, as shown in Figure 1.By comparing the same elements in different artificial intelligence applications and looking for the relevance, different applications can be combined to produce a factor affecting multi-application feedback to produce multi-sided decision-making, resulting in the effect of 1 + 1 > 2.
The business scenarios of the new power system influence each other, and it is difficult to reflect the overall relevance of the system only from their respective business scenarios.Therefore, the application of complex adaptive system theory to analyze the technical application of the new power system can fully show the influence of artificial intelligence, and face the reality of the application of artificial intelligence in the new electric power system.

Theoretical Application of Artificial Intelligence technology in New Power System
With the gradual implementation of China's "Double Carbon" target, the energy structure is gradually shifting to multi-source clean energy.The new energy access to the network has many characteristics, such as uncertainty, instability, spatial and temporal differences, multi-energy coupling and so on, which make the power system structure more complex.Artificial intelligence with advanced computer technology and data analysis, anthropomorphic learning has changed the traditional analysis logic and promoted the safe and reliable development of power system and enhanced economic benefits.
1) Technology breakthrough and application theory.We will strengthen theoretical research on basic technologies such as Big Data Intelligence, cross-media perceptual computing, human-computer hybrid intelligence, group intelligence, autonomous collaboration and decision-making, and form a hybrid enhanced intelligence system, which is close to human intelligence to be applied to major business scenarios such as construction, operation, inspection, repair and service in the new power system, so as to realize the innovative theoretical system of autonomous intelligent unmanned power system.
2) Layout frontier business theory.In view of the directions that may lead to changes in human intelligence, such as ChatGPT, we need to strengthen the research on cross-domain technologies such as advanced machine learning and brain-like intelligent computing, and establish a hybrid model of efficient computing and quantum algorithms to form a new quantum artificial intelligence power system architecture with efficient, accurate and autonomous power system operation.

Integration of Artificial Intelligence Technology and New Power System
Faced with the challenge of unstable supply of new energy, unstable operation of power grid and insufficient carrying capacity of power equipment, the development of power grid tends to be intelligent, which requires the application of more efficient and reliable power system technology.Artificial intelligence technology can realize the accurate mapping of artificial intelligence to the new power system and improve the stability of system operation through topology data integration and statistics of data standards in each link of source network load storage.The application analysis is shown in Figure 2. Based on the comprehensive analysis of the energy data, power grid conditions and consumption market conditions of new energy [6] , relevant control measures are taken to connect new energy to the grid after appropriate adjustment and coordinate consumption with traditional energy.At the same time, with the development of engineering intelligent tools and high-tech energy equipment, the stability of electronic equipment operation is improved, and the carrying capacity of equipment is strengthened.

Application scenario analysis of Artificial Intelligence in New Power System
Depending on the important basis of AI's innovative algorithm and computing power, each business scenario is connected in series and integrated into the whole new power system, the technical requirements of AI are analyzed for each business scenario, the computing architecture and management rules matching business scenarios are constructed, and the intelligent devices suitable for different business scenarios are developed.Relevant implementation means such as regulation, control, maintenance and prediction are applied to the business scenario to realize the intelligent result presentation of the business scenario and form the integrated intelligent comprehensive application framework of the new power system, as shown in Figure 3. wind power, sunshine, weather forecast, area, equipment forecast combined with power conversion, power market economy, construction difficulties, electricity demand analysis and to estimate the risk of power station construction can explore the demand of power station construction and the best place to build new energy power stations, which can maximize the use of natural resources.Considering the construction cost and expected risk of multi-agent transmission and distribution construction planning, the optimal power grid construction scheme is determined by using swarm intelligence method.
2) Construction scenario.The power generation side and transmission side of the new power system have a large area and a large amount of work, and some areas are located in remote areas, resulting in a long construction period and a large demand for labors.Research and development of intelligent robots for power grid construction, with assembly building design.The robots will undertake construction, paint spraying, loading and other work, which will dangerous work and heavy physical work workers, increase the cooperation between construction robots and robots, reduce the risk of falling from high altitude, electric shock and other work, while reducing work errors, shortening the construction period, reducing the cost of labor and materials and breaking through the construction difficulties.
3) Equipment safety monitoring scenarios.By collecting equipment image information, identifying the running state of equipment and conducting risk assessment, the safety of equipment can be effectively guaranteed.It can identify and monitor equipment quality automatically, use trajectory, health status, life cycle and other information, provide early warning of risk sources, timely and targeted management of equipment, and reduce the probability of equipment and power grid operation accidents.By tracking the whole process of the equipment, and analyzing the health status of the equipment, can ensure the safe use of the power equipment [7] .
4) Power fault diagnosis scenario.Artificial intelligence deep learning technology is used to identify the intrinsic coupling characteristics of large power data and reduce the complexity of power failure.Combination of expert system method, rough set method and artificial neural network method can deal with a variety of different power failure situations.According to the different conditions of data integrity and data accuracy, the appropriate diagnosis method is selected independently.Combing with anthropomorphic learning technology, to form the autonomous learning path, and the diagnosis database is supplemented to adapt to the diagnosis situation of large-scale power system and improve the accuracy of power fault diagnosis.
5) Virtual exercise scenario.In recent years, natural disasters have occurred frequently, and construction facilities around the country have been damaged in different degrees by disasters.Through the establishment of intelligent models of virtual power stations and power grids, the impact of extreme weather such as storms, earthquakes and thunderstorms on each end of the source network load storage is predicted, and the disaster situation and results are simulated, so as to realize the organic combination and efficient interaction between virtual reality, augmented reality and artificial intelligence, which can help predict and prevent the impact of disasters, improve the ability to resist disasters, and minimize the loss of disaster.
6) Power grid intelligent dispatching scenario.Through the construction of a comprehensive intelligent dispatch center for multi-energy grid access, we can coordinate power supply, power grid, and load factors globally, and use artificial intelligence technology to balance the volatility and complexity of multi-energy grid access, and complete the multi-scenario dispatch center analysis predict, as shown in Figure 4.An intelligent inspection robot with autonomous unmanned system is invented for substations, which is equipped with technologies such as complex dynamic scene perception and understanding, real-time precise positioning, adaptive intelligent navigation for complex environment, intelligent detection of equipment performance and analysis of fault conditions for recording and feedback.With intelligent patrol UAV, image comparison and analysis are fully applied to achieve full link coverage of patrol inspection. [8]) User service scenario.By using advanced network technology and sensor technology for remote real-time meter data acquisition, the meter information is classified and collated.At the same time, intelligent user service stations are established, and intelligent customer service centers are developed to support remote autonomous power services for power users, including but not limited to precise services such as installation, maintenance, inquiry and payment.Continuously improve product quality and intelligent level, and provide effective solutions for the construction of intelligent residential areas.

Summary and recommendations
With the innovative research and development of artificial intelligence and the accumulation of power grid data in the construction of new power system, a large number of studies have been carried out around the application of artificial intelligence technology in smart grid.Through artificial intelligence technology, the new power system is more intelligent and autonomous, so as to achieve the goals of stable operation, problem prediction, autonomous maintenance and intelligent regulation.At present, artificial intelligence is still in the research stage of underlying technology, and large-scale application is still facing technical bottlenecks.In order to accelerate the application of artificial intelligence in new power system scenarios, this paper puts forward relevant suggestions from the following aspects.
1) Platform construction: building an intelligent power system platform.Artificial intelligence in the new power system is developing towards the integration of hybrid technologies, building an intelligent power system platform that integrates generation, transmission, distribution, transformation, power consumption, perception and prediction, overall management and control, and safety maintenance, breaking data barriers and to provide a platform basis for the learning and growth of artificial intelligence.
2) Technical algorithm: improving the idea of combining numbers and objects and breaking through technical difficulties.In view of the stability of artificial intelligence computing in the new power system, it is necessary to explore the breakthrough points, find out the correlation points between the physical environment and the data environment, improve the idea of combining numbers and objects, and apply the correlation between data governance and physical models to the new generation of artificial intelligence power system, so as to achieve the harmonious combination of numbers and objects truly, and overcome the technical bottleneck.
3) System management: Establishing a new artificial intelligence management system for power system.The artificial intelligence technology, intelligent system and intelligent machine used in the new power system should be properly controlled, and a complete management system should be established to avoid the risks of use safety and information leakage, so as to achieve the effect of full use and make the artificial intelligence more safe and reliable in the new power system.4) Application evaluation: Establishing the evaluation system of artificial intelligence applied to the new power system.It is of great significance through qualitative and quantitative evaluation to test whether the application of artificial intelligence in the new power system is scientific and feasible, to judge the applicability, practicability and feasibility of the application scenarios, and to ensure that the designed application scenarios can be truly implemented.

Figure 1 .
Figure 1.Illustration of complex adaptive system theory

Figure 2 .
Figure 2.Schematic diagram of analysis of artificial intelligence technology in new power system

Figure 3 .
Figure 3. Schematic Diagram of Intelligent Integrated Application Framework of New Power System 1) New energy generation side addressing scenario.Through artificial intelligence to calculate the

Figure 4 .
Figure 4. Schematic diagram of multi-energy grid dispatching 7) Maintenance scenario.An intelligent inspection robot with autonomous unmanned system is invented for substations, which is equipped with technologies such as complex dynamic scene perception and understanding, real-time precise positioning, adaptive intelligent navigation for complex environment, intelligent detection of equipment performance and analysis of fault conditions for recording and feedback.With intelligent patrol UAV, image comparison and analysis are fully applied to achieve full link coverage of patrol inspection.[8]8) User service scenario.By using advanced network technology and sensor technology for remote real-time meter data acquisition, the meter information is classified and collated.At the same time, intelligent user service stations are established, and intelligent customer service centers are developed to support remote autonomous power services for power users, including but not limited to precise services such as installation, maintenance, inquiry and payment.Continuously improve product quality and intelligent level, and provide effective solutions for the construction of intelligent residential areas.

Table 1 .
for details.Application Results of Artificial Intelligence in Energy IndustryArtificial intelligence enhanced software platforms and intelligent tools can mine energy data to predict potential energy loss and provide automation support.