Multi-agent-based power system optimization scheduling system and method

Abstract

The invention belongs to the technical field of power system optimization scheduling, and particularly relates to a multi-agent-based power system optimization scheduling system and method. The multi-agent-based electric power system is electrically connected with a market evaluation layer, an electric power parameter prediction and management layer, a centralized decision-making layer, a distributed control layer and an equipment monitoring and management layer through wires, and the market evaluation layer is bidirectionally and electrically connected with the electric power parameter prediction and management layer through a wire. The electric power parameter prediction and management layer is bidirectionally and electrically connected with the centralized decision-making layer through a wire, the centralized decision-making layer is bidirectionally and electrically connected with the distributed control layer through a wire, the distributed control layer is bidirectionally and electrically connected with the equipment monitoring and management layer through a wire, and information interaction with an intelligent power distribution station is completed through an information bus. According to the system, the optimal scheduling of the power system is divided into several layers, so that the coupling degree of modules in the power system is greatly reduced, and the system can coordinate and call related resources to the greatest extent.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power system optimal dispatching, and particularly relates to a power system optimal dispatching system and method based on multiple agents. BACKGROUND

[0002] The proportion of non-fossil energy on the power supply side in the new power system is very high. Due to the intermittent, random and volatile characteristics of these renewable energy sources, it brings certain problems to the safe and stable operation of the power system. In order to solve this problem, many scholars have designed and researched the source-grid-load-storage integrated multi-element coordinated control interaction system.

[0003] The application with the publication number CN 115566735 A discloses an optimal dispatching method, device and system for a power system, which belongs to the technical field of electrical engineering. The method comprises the following steps: based on the Bellman optimality principle, a centralized optimal dispatching model of the power system is established by using the technical parameters of each element in the power system; based on the Lagrange equation and the consistency theory, the centralized optimal dispatching model is split into sub-optimal dispatching models corresponding to each element; a power limitation factor is applied to each sub-optimal dispatching model to improve each sub-optimal dispatching model; a power system consistency distributed algorithm containing a first-order cost function element is used to solve each improved sub-optimal dispatching model to obtain a convergent optimal solution; and a distributed optimal dispatching scheme corresponding to the convergent optimal solution is used to implement dispatching. The application can overcome the convergence instability problem of the traditional consistency optimization algorithm when optimizing the power system containing the first-order cost function element, and ensure the feasibility and optimality of the solution.

[0004] The application with the publication number CN 112467807 B relates to a multi-energy power system day-ahead optimal dispatching method and system. The method comprises the following steps: generating a wind power and photovoltaic power generation power scene based on an improved generative adversarial network of Wasserstein distance; defining a virtual net load and generating a virtual net load according to the wind power and photovoltaic power generation power scene; based on the virtual net load, establishing a day-ahead optimal dispatching model of the multi-energy power system; and using an improved adaptive genetic algorithm to solve the day-ahead optimal dispatching model of the multi-energy power system to obtain a final dispatching result. The application can optimize the dispatching of the multi-energy power system containing wind, light, water and fire storage, reduce the adverse effects on the safe operation of the power system caused by the uncertainty and volatility of wind power and photovoltaic power generation, and improve the consumption level of renewable energy.

[0005] However, most of these studies stay on the interaction of source-source, source-network, network-load-storage, and few scholars can fully consider the four elements of source, network, load and storage, and the scheduling strategy based on the four elements has not been fully studied. And the new type of power system in the future may introduce more types of renewable energy with a larger proportion, and the uncertainty and volatility of the load end will also increase with the development of society. And the coupling degree between the subsystems in the existing optimization scheduling system is too high, and the autonomy of the subsystem is insufficient, and it does not have the ability to solve the problem alone, so once a module appears abnormal, the system cannot operate normally, which will cause great waste of resources, and also cause damage to the non-fault module, and for the occurrence of faults and natural disasters and other emergencies, the system responds too slowly and cannot form a scheduling scheme in time. Therefore, an optimization scheduling method that considers the four elements is urgently needed to cope with greater risks. SUMMARY

[0006] The purpose of the present application is to provide a multi-agent-based power storage system optimization scheduling system and method that addresses the problems in the prior art. The system applies multi-agent technology to the optimization scheduling of the power system, divides the optimization scheduling of the power system into several layers, each layer being a large agent, greatly reducing the coupling degree of the modules in the power system, and enabling the system to maximize the coordination of related resources.

[0007] The technical solution of the present application is:

[0008] The multi-agent-based power system optimization scheduling system is electrically connected with the market evaluation layer, the power parameter prediction and management layer, the centralized decision-making layer, the distributed control layer and the equipment monitoring and management layer through wires, the market evaluation layer and the power parameter prediction and management layer are bidirectionally electrically connected through wires, the power parameter prediction and management layer and the centralized decision-making layer are bidirectionally electrically connected through wires, the centralized decision-making layer and the distributed control layer are bidirectionally electrically connected through wires, the distributed control layer and the equipment monitoring and management layer are bidirectionally electrically connected through wires, and the market evaluation layer, the power parameter prediction and management layer, the centralized decision-making layer, the distributed control layer and the equipment monitoring and management layer are all agents, and information interaction is completed with the intelligent power distribution station through an information bus.

[0009] Specifically, the method for using the multi-agent-based power system optimization scheduling system of claim 1 comprises the following steps:

[0010] S1, the market evaluation layer agent evaluates the operation cost of the power system, and predicts the demand and transaction of the market to provide economic indicators for the optimization scheduling of the system;

[0011] S2, the power parameter prediction and management layer agent predicts the power load, estimates the power margin required to compensate for the impact of the accident, and collects the state of the adjustable load to provide power indicators for the optimized dispatch of the system;

[0012] S3, the centralized decision-making layer agent collects various indicator information, and according to the different optimization functions, different models are established, and then the optimal dispatching scheme is given;

[0013] S4, the distributed control layer agent utilizes the complementary characteristics and regulation and control capabilities of resources at all levels to coordinate various resources, and through optimized dispatching, reduces the operating cost of the power grid and the probability of occurrence of a large-scale power failure;

[0014] S5, the device monitoring and management layer agent monitors the device and gives an early warning for the abnormality, and manages the load to provide device state information for the optimized dispatch of the system.

[0015] Specifically, the evaluation and prediction of the market evaluation layer agent in step S1 specifically includes the following steps:

[0016] S1-1: Generation cost prediction: estimate the comprehensive generation cost by combining historical generation cost data, current fossil fuel sales price, system operation and maintenance cost and depreciation, and policy influencing factors;

[0017] S1-2: Market demand and transaction prediction: predict market demand and transactions by combining historical data, changes in market conditions, peak and valley occurrence time, seasonal alternation, and generation cost factors, and also make a preliminary prediction of the price of electricity purchase;

[0018] S1-3: Resource allocation and planning: based on demand, supply, and cost, market, evaluate the economic efficiency of fossil fuel power generation and new energy power generation to facilitate resource allocation and planning, and provide economic indicators for the final optimized dispatching instruction;

[0019] S1-4: Market supervision and risk management: real-time monitor the changes in the market environment, and predict and manage the possible risks, and respond in a timely manner to changes in market and user demand, and update the relevant data.

[0020] Specifically, the power load prediction and estimation of the power parameter prediction and management layer agent in step S2 specifically includes the following steps:

[0021] S2-1: Power load prediction: use neural network algorithm to predict the power load of the distribution network according to the historical data of the user power load;

[0022] S2-2: Run risk assessment and corresponding margin calculation: when the power grid fails, according to the historical data caused by the accident, combined with the current equipment running state, the risk state is evaluated, and the corresponding margin is calculated, which provides a guide for the management of subsequent energy storage devices and adjustable load;

[0023] S2-3: Distributed energy generation equipment output prediction: using historical data and related weather information, the available capacity of distributed photovoltaic power and other power sources is evaluated in a range, to ensure that clean energy can be preferentially consumed;

[0024] S2-4: Collection of adjustable load state: according to the current power system operating condition and historical data, the adjustable load condition is obtained, and the adjustable capacity required by the corresponding power system is compared with the actual data, to provide a reference for subsequent optimization scheduling.

[0025] Specifically, the specific steps of the central decision layer agent collecting each index information and giving the optimal scheduling scheme according to different optimization functions in step S3 are as follows:

[0026] S3-1: Data collection and processing: the data processed by other layers and the historical data of system operation are sorted and summarized, to lay a good foundation for the establishment of subsequent models;

[0027] S3-2: Establishment of optimization function and modeling of system: according to the target to be achieved, the corresponding optimization function is established by comprehensively considering economic benefit and environmental benefit and other factors, and it is ensured that the operation of the power system meets the basic constraints: distribution network safety constraint: U i,min ≤U i ≤U i,max ,I ij,min ≤I ij ≤I ij,max ; Distribution network frequency constraint: f min f f max ; Energy storage battery constraint: Distributed power constraint: P DGimin P DGi P DGimax ; Wherein U i,min , U i , U i,max respectively represent the minimum voltage, actual voltage and maximum voltage allowed for node i when the distribution network system is normally operated; I ij,min , I ij , I ij,max respectively represent the minimum allowable current, actual current and maximum allowable current on branch ij when the distribution network system is normally operated; f min , f, f maxrespectively represent the minimum frequency allowed, the actual frequency, the maximum frequency when the system is running normally; P cha , P dis respectively represent the charging power and the discharging power of the energy storage battery; P DGi , P DGimin , P DGimax respectively represent the power generation of the i-th distributed power supply, the minimum allowed power generation, and the maximum allowed power generation;

[0028] S3-3: Training and optimization of the model: based on historical data, train the basic neural network model in advance, and then adjust the corresponding framework according to the changed target function and input indicators, and train and optimize it;

[0029] S3-4: System integration test: debug and optimize the trained network and the constructed model;

[0030] S3-5: Feedback and optimization: refer to the generated results to optimize the scheduling scheme, and feed back the scheduling scheme to each intelligent body to optimize and adjust the resources of the system.

[0031] Specifically, the distribution control layer intelligent agent utilizes the complementary characteristics and regulation and control capabilities of resources at each level in step S4, and the specific steps are as follows:

[0032] S4-1: Distribution control: master the source, network, load, and storage resource status at each level, compare the difference between the system required and the system adjustable capacity, and form a specific adjustment control strategy for each distributed energy source;

[0033] S4-2: Intelligent agent decision: overall plan the strategies provided by other sub-intelligent agents at this level, and issue specific regulation and control instructions;

[0034] S4-3: Communication and cooperation, evaluation and optimization: the distribution control intelligent agent functions to coordinate various resources, needs to interact with other modules, and timely evaluates and optimizes the resource conflicts and problems caused by optimization and coordination, and transmits the information to the centralized decision-making layer.

[0035] Specifically, the device monitoring and management layer intelligent agent monitors the devices and gives early warning for the abnormalities in step S5, including the following steps:

[0036] S5-1: Device monitoring and data transmission and storage: the device layer is directly connected with power distribution equipment, power generation equipment, and energy storage equipment, monitors these devices through Internet of Things technology, obtains their running states in real time, and completes information interaction with other layers through the information bus by transmitting the obtained data;

[0037] S5-2: Abnormal early warning: based on the operation data of the monitored equipment, it is compared with the data of the normal operation state, whether the equipment has a risk of failure is judged and predicted, whether the equipment is in a failure state, if an abnormal state occurs, information is fed back to each layer in time, and through the load management and optimization agent of the same layer, corresponding processing measures are put forward;

[0038] S5-3: Load management and equipment performance optimization: this sub-agent directly contacts each equipment, manages and optimizes the performance of the equipment according to the given optimization scheduling scheme, and feeds back the results to other agents.

[0039] The beneficial effects of the present application are: compared with the power system optimization scheduling technology in the prior art, the multi-agent technology is introduced in the present application, the autonomy, interaction and intelligence of the multi-agent system are fully utilized, the functions of each agent in the planning system are planned, the communication interaction between the agents is coordinated, complex problems can be quickly and flexibly solved, a new idea and mode are provided for the scheduling system, and each level of the system is a large agent, the large agents are composed of many sub-agents, each sub-agent bears a part of the function, the coupling degree between the agents is greatly reduced, and the functions of other agents can still be executed when some modules fail, so that the loss is minimized.

[0040] The present application fully considers the four elements of source, network, load and storage, and truly realizes the coordination and interaction of source, network, load and storage, which provides a better solution for the volatility problem caused by the integration of more types and larger proportions of new energy into the power grid in the future. The system provided by the present application also has a special market evaluation layer, fully considers the economic factors, can predict the demand and transaction of the market, makes corresponding adjustment according to different demand response of users and market, and then reduces the production cost, improves the competitiveness of products, and ensures the economic benefit of system operation. The system provided by the present application realizes the connection between different levels through the information bus, and the information can be conveyed to the intelligent power distribution station through the information bus. When the intelligent power distribution station determines the optimization scheduling scheme, the intelligent power distribution station feeds back to each layer through the information bus, so that the work of each level is more stable. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a system framework structure schematic diagram of the present application;

[0042] Figure 2 is a method principle flow chart of the present application;

[0043] Figure 3 is a market evaluation agent work flow chart;

[0044] Figure 4 This is a flowchart of the workflow for an intelligent agent for predicting and managing power parameters;

[0045] Figure 5 This is a flowchart of the workflow of a centralized decision-making intelligent agent;

[0046] Figure 6 This is a flowchart of the distributed control agent workflow;

[0047] Figure 7 This is a flowchart of the intelligent agent for equipment monitoring and management. Detailed Implementation

[0048] The technical solution of the present invention will be described in detail below with reference to specific embodiments.

[0049] Example 1

[0050] like Figure 1 The diagram shown is a schematic of the framework structure of the power system optimization and dispatching system based on multi-agent provided in this embodiment. The power system based on multi-agent is electrically connected to the market assessment layer, the power parameter prediction and management layer, the centralized decision-making layer, the distributed control layer, and the equipment monitoring and management layer via conductors. The market assessment layer and the power parameter prediction and management layer are connected bidirectionally via conductors. The power parameter prediction and management layer and the centralized decision-making layer are connected bidirectionally via conductors. The centralized decision-making layer and the distributed control layer are connected bidirectionally via conductors. The distributed control layer and the equipment monitoring and management layer are connected bidirectionally via conductors. The market assessment layer, the power parameter prediction and management layer, the centralized decision-making layer, the distributed control layer, and the equipment monitoring and management layer are all intelligent agents, and they interact with the smart distribution station through an information bus.

[0051] Example 2

[0052] like Figure 2 The diagram shown is a flowchart illustrating the principle of the power system optimization scheduling system based on multi-agent technology provided in this embodiment. The method specifically includes the following steps:

[0053] S1. The market assessment layer intelligent agent assesses the operating costs of the power system and predicts market demand and transactions, providing economic indicators for the optimized scheduling of the system.

[0054] S2. The power parameter prediction and management layer intelligent agent predicts the power load, estimates the power margin required to compensate for the impact of the accident, and collects the status of adjustable load to provide power indicators for the system's optimized scheduling.

[0055] S3. The centralized decision-making layer intelligent agent collects information on various indicators, establishes different models according to different optimization functions, and then provides the optimal scheduling scheme.

[0056] S4, the distribution control layer agent utilizes the complementary characteristics and regulation capacity of resource net load storage resources at each level, coordinates various resources, reduces power grid operation cost through optimized scheduling, and reduces the probability of occurrence of large-scale power failure accidents;

[0057] S5, the device monitoring and management layer agent monitors the device, gives an early warning for the occurrence of an abnormality, and manages the load to provide device state information for the optimized scheduling of the system.

[0058] Embodiment 3

[0059] The embodiment provides detailed operation steps of the evaluation and prediction of the market evaluation layer agent in step S1 in embodiment 2, and includes the following steps.

[0060] S1-1: Generation cost prediction: estimate the comprehensive generation cost by combining historical generation cost data, current fossil fuel sales price, system operation and maintenance cost and depreciation, and policy influencing factors;

[0061] S1-2: Market demand and transaction prediction: predict the market demand and transaction by combining historical data, market situation changes, peak and valley occurrence time, seasonal alternation, and generation cost factors, and also can preliminarily predict the electricity purchase price;

[0062] S1-3: Resource allocation and planning: based on demand, supply, and cost, market, evaluate the economy of fossil fuel power generation and new energy power generation, facilitate resource allocation and planning, and provide economic indicators for the final optimized scheduling instruction;

[0063] S1-4: Market supervision and risk management: real-time monitor the changes of market environment, and predict and manage the possible risks, and can respond in time to the changes of market and user demand, and update the related data.

[0064] Embodiment 4

[0065] The embodiment provides specific steps of the power load prediction and estimation of the power parameter prediction and management layer agent in step S2 in embodiment 2, and includes the following steps.

[0066] S2-1: Power load prediction: predict the power load of the distribution network by using a neural network algorithm according to the historical data of the user power load;

[0067] S2-2: Run risk assessment and corresponding margin calculation: when the power grid fails, according to the historical data of the impact caused by the accident, combined with the current equipment operation state, the risk state is evaluated, and the corresponding margin is calculated, which provides a guide for the management of subsequent energy storage devices and adjustable load;

[0068] S2-3: Distributed energy generation device output prediction: using historical data and related weather information, the available capacity of distributed photovoltaic power and other power sources is evaluated within a certain range, to ensure that clean energy can be preferentially consumed;

[0069] S2-4: Collection of adjustable load state: according to the current power system operating condition and historical data, the adjustable load condition is obtained, and the adjustable capacity required by the corresponding power system is compared with the actual data, to provide a reference for subsequent optimization scheduling.

[0070] Embodiment 5

[0071] The embodiment provides specific steps of the central decision layer agent collecting index information in step S3 described in embodiment 2, and establishing different models according to different optimization functions, and then giving an optimal scheduling scheme as follows:

[0072] S3-1: Data collection and processing: the data processed by other layers and the historical data of system operation are sorted and summarized, to lay a good foundation for subsequent model establishment;

[0073] S3-2: Establishment of optimization function and modeling of system: according to the target to be achieved, the corresponding optimization function is established by comprehensively considering economic benefit and environmental benefit and other factors, and it is ensured that the operation of the power system meets the basic constraints: distribution network safety constraint: U i,min ≤U i ≤U i,max ,I ij,min ≤I ij ≤I ij,max ; distribution network frequency constraint: f min f f max ; energy storage battery constraint: distributed power constraint: P DGimin P DGi P DGimax ; wherein U i,min , U i , U i,max respectively represent the minimum voltage, actual voltage and maximum voltage allowed for node i when the distribution network system is normally operated; I ij,min , I ij , I ij,max respectively represent the minimum allowable current, actual current and maximum allowable current on branch ij when the distribution network system is normally operated; fmin f, f max respectively represent the minimum frequency allowed when the system is running normally, the actual frequency, the maximum frequency; P cha P, P dis respectively represent the charging power and the discharging power of the energy storage battery; P DGi P, P DGimin P, P DGimax respectively represent the power generation of the i-th distributed power supply, the minimum allowed power generation, and the maximum allowed power generation;

[0074] S3-3: Training and optimization of the model: based on historical data, a basic neural network model is trained in advance, and then according to the changed target function and input index, the corresponding framework is adjusted and trained and optimized;

[0075] S3-4: System integration test: debug and optimize the trained network and the constructed model;

[0076] S3-5: Feedback and optimization: refer to the generated results to optimize the formulation of the scheduling scheme, and feed back the scheduling scheme to each intelligent body to optimize and adjust the resources of the system.

[0077] Embodiment 6

[0078] The embodiment provides the specific steps of the distributed control layer intelligent agent in step S4 of the method provided in embodiment 2, which utilizes the complementary characteristics and regulation and control capabilities of the resources at each level to coordinate the resources as follows:

[0079] S4-1: Distributed control: master the source, network, and load storage resource status at each level, compare the difference between the system required and the system adjustable capacity, and form a specific adjustment and control strategy for each distributed energy source;

[0080] S4-2: Intelligent agent decision: overall plan the strategies provided by other sub-intelligent agents at this level, and issue specific regulation and control instructions;

[0081] S4-3: Communication and cooperation, evaluation and optimization: the distributed control intelligent agent functions to coordinate the resources, needs to interact with other modules, and timely evaluates and optimizes the resource conflicts and problems caused in the optimization and coordination, and transmits the information to the centralized decision-making layer.

[0082] Embodiment 7

[0083] The embodiment provides the steps of the equipment monitoring and management layer intelligent agent monitoring the equipment and warning the abnormality in step S5 of the method provided in embodiment 2, including the following steps:

[0084] S5-1: Equipment monitoring and data transmission and storage: the equipment layer is directly connected with power distribution equipment, power generation equipment and energy storage equipment, and the equipment is monitored through Internet of Things technology, the running state of the equipment is obtained in real time, and the obtained data is transmitted to other layers through an information bus to complete information interaction;

[0085] S5-2: Abnormal early warning: based on the monitored equipment operation data, the data is compared with the normal operation state data, whether the equipment has a risk of failure, whether the equipment is in a failure state is judged and predicted, if an abnormal state occurs, information is fed back to each layer in time, and through the load management and optimization agent of the same layer, corresponding processing measures are proposed;

[0086] S5-3: Load management and equipment performance optimization: this sub-agent is directly connected with each equipment, manages and optimizes the performance of the equipment according to the given optimization scheduling scheme, and feeds back the result to other agents.

[0087] The application provides a system that applies multi-agent technology to a power system, and refines the functions of the system, each function is undertaken by a corresponding sub-agent, which greatly reduces the coupling degree between different levels, so that when some modules are abnormal, other modules can still operate normally, resources are fully utilized, and the four elements of source, network, load and storage are comprehensively considered, the original limitation of source-source, source-network, network-load-storage is broken through, and integrated optimization scheduling of source-network-load-storage is realized.

[0088] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application and not to limit them; although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or some technical features can be replaced by equivalent; without departing from the spirit of the technical solutions of the present application, they should be covered in the technical solution range of the present application claimed.

Claims

1. A power system optimization dispatching system based on multi-agent systems, characterized in that, The multi-agent-based power system is electrically connected to the market assessment layer, power parameter prediction and management layer, centralized decision-making layer, distributed control layer, and equipment monitoring and management layer via wires. The market assessment layer and the power parameter prediction and management layer are bidirectionally connected via wires. The power parameter prediction and management layer and the centralized decision-making layer are bidirectionally connected via wires. The centralized decision-making layer and the distributed control layer are bidirectionally connected via wires. The distributed control layer and the equipment monitoring and management layer are bidirectionally connected via wires. Each of the market assessment layer, power parameter prediction and management layer, centralized decision-making layer, distributed control layer, and equipment monitoring and management layer is an intelligent agent, and they interact with the intelligent distribution station through an information bus.

2. The method for optimizing and dispatching a power system based on a multi-agent system as described in claim 1, characterized in that, Includes the following steps: S1. The market assessment layer intelligent agent assesses the operating costs of the power system and predicts market demand and transactions, providing economic indicators for the optimized scheduling of the system. S2. The power parameter prediction and management layer intelligent agent predicts the power load, estimates the power margin required to compensate for the impact of the accident, and collects the status of adjustable load to provide power indicators for the system's optimized scheduling. S3. The centralized decision-making layer intelligent agent collects information on various indicators, establishes different models according to different optimization functions, and then provides the optimal scheduling scheme. S4. The distributed control layer intelligent agent utilizes the complementary characteristics and regulation capabilities of the resource network load and storage resources at all levels to coordinate various resources, reduce power grid operating costs, and reduce the probability of major power outages through optimized scheduling. S5. The device monitoring and management layer intelligent agent monitors the device, issues early warnings for any anomalies, and manages the load, providing device status information for the system's optimized scheduling.

3. The power system optimization scheduling method based on multi-agent systems according to claim 2, characterized in that, The evaluation and prediction by the market evaluation layer agent in step S1 specifically includes the following steps: S1-1: Power generation cost forecast: Estimate the overall power generation cost by combining historical power generation cost data, current fossil fuel prices, system operation and maintenance costs and depreciation, and policy influencing factors; S1-2: Market Demand and Transaction Forecast: This forecast combines historical data, changes in market conditions, the timing of peak and trough events, seasonal changes, and power generation cost factors to predict market demand and transactions. It can also provide a preliminary forecast of electricity purchase prices. S1-3: Resource Allocation and Planning: Based on demand, supply, cost, and market, assess the economics of fossil fuel power generation and new energy power generation to facilitate resource allocation and planning, and provide economic indicators for the final optimized dispatch instructions. S1-4: Market Surveillance and Risk Management: Monitor changes in the market environment in real time, predict and manage potential risks, respond promptly to changes in market and user needs, and update relevant data.

4. The power system optimization scheduling method based on multi-agent systems according to claim 2, characterized in that, Step S2, which involves power parameter prediction and the management layer agent predicting and estimating power load, specifically includes the following steps: S2-1: Power load forecasting: Based on historical user power load data, a neural network algorithm is used to forecast the power load of the distribution network; S2-2: Operational Risk Assessment and Corresponding Margin Calculation: When a power grid failure occurs, the risk status is assessed based on historical data of the impact of the accident and the current equipment operating status, and the corresponding margin is calculated to provide guidance for the subsequent management of energy storage equipment and adjustable loads. S2-3: Output prediction of distributed energy generation equipment: Using historical data and relevant meteorological information, a range of power generation capacity of distributed photovoltaic and other power sources is assessed to ensure that clean energy can be prioritized for consumption. S2-4: Adjustable load status acquisition: Based on the current power system operating status and historical data, the adjustable load status is obtained, and the corresponding adjustable capacity required by the power system is compared with the actual data to provide a reference for subsequent optimized scheduling.

5. The power system optimization scheduling method based on multi-agent systems according to claim 1, characterized in that, In step S3, the centralized decision-making agent collects information on various indicators, establishes different models based on different optimization functions, and then provides the specific steps for the optimal scheduling scheme as follows: S3-1: Data Acquisition and Processing: Organize and summarize the data processed by other layers and the historical data of system operation to lay the groundwork for the establishment of subsequent models; S3-2: Establishment of Optimization Function and System Modeling: Based on the objectives to be achieved, establish the corresponding optimization function while comprehensively considering economic and environmental benefits, and ensure that the operation of the power system meets basic constraints: Distribution network security constraints: U i,min ≤U i ≤U i,max ,I ij,min ≤I ij ≤I ij,max Distribution network frequency constraints: f min ff max Energy storage battery constraints: Distributed power source constraints: P DGimin P DGi P DGimax ;where U i,min U i U i,max These represent the minimum allowable voltage, actual voltage, and maximum voltage of node i during normal operation of the distribution network system, respectively. I ij,min I ij I ij,max These represent the minimum allowable current, actual current, and maximum allowable current on branch ij, respectively, during normal operation of the distribution network system; f min f, f max These represent the minimum allowed frequency, actual frequency, and maximum frequency during normal system operation, respectively; P cha P dis These represent the charging power and discharging power of the energy storage battery, respectively; P DGi P DGimin P DGimax Let represent the power generation capacity, minimum allowable power generation capacity, and maximum allowable power generation capacity of the i-th distributed power source, respectively. S3-3: Model Training and Optimization: Based on historical data, a basic neural network model is trained in advance, and then the framework is adjusted according to the modified objective function and input indicators, and then trained and optimized. S3-4: System integration testing: Debugging and optimizing the trained network and the constructed model; S3-5: Feedback and Optimization: Based on the results generated, formulate an optimized scheduling plan and feed the plan back to each intelligent agent to optimize and adjust various resources of the system.

6. The power system optimization scheduling method based on multi-agent systems according to claim 2, characterized in that, In step S4, the distributed control layer agent utilizes the complementary characteristics and regulation capabilities of resource network load storage resources at all levels to coordinate various resources. The specific steps are as follows: S4-1: Distributed control: Understand the resource status of source, grid, load and storage at each level, compare the difference between the system's required capacity and the system's adjustable capacity, and formulate targeted regulation and control strategies for each distributed energy source. S4-2: Agent Decision-Making: Coordinate the strategies provided by other sub-agents in this layer and issue specific control instructions; S4-3: Communication and Collaboration, Evaluation and Optimization: The distributed control agent plays the role of coordinating various resources. It needs to interact with other modules and promptly evaluate and optimize resource conflicts caused by optimization and coordination, and transmit the information to the centralized decision-making layer.

7. The power system optimization scheduling method based on multi-agent systems according to claim 2, characterized in that, The device monitoring and management layer intelligent agent in step S5 monitors the device and issues early warnings for any anomalies, including the following steps: S5-1: Equipment monitoring and data transmission and storage: The equipment layer is directly connected to power distribution equipment, power generation equipment and energy storage equipment. It monitors these devices through Internet of Things technology, obtains their operating status in real time, and exchanges the obtained data with other layers through the information bus. S5-2: Anomaly Warning: Based on the monitored equipment operation data, it compares the data with the normal operation data to determine and predict whether the equipment is at risk of failure and whether it is already in a fault state. If an abnormal state occurs, the information is promptly fed back to each layer, and corresponding handling measures are proposed through the load management and optimization agent of the same layer. S5-3: Load Management and Equipment Performance Optimization: This sub-agent communicates directly with each device, manages the device and optimizes its performance according to the given optimization scheduling scheme, and feeds back the results to other agents.

Images