A power grid dispatching fault intelligent disposal method based on multi-agent cooperative operation

Abstract

The application provides a power grid scheduling fault intelligent disposal method based on multi-agent collaborative operation, relates to the technical field of power systems, and comprises the following steps: acquiring current operation state information of a power grid, and performing multi-dimensional data fusion on the current operation state information to construct a power grid state space; performing fault risk calculation according to the multi-dimensional data fusion result; constructing a multi-agent Markov decision process model by using the fault risk calculation result; introducing an attention mechanism based on the multi-agent Markov decision process model to perform dynamic feature aggregation; calculating resource dynamic scheduling according to the dynamic feature aggregation result, and performing fault processing according to the scheduling result; and the application solves the weakness of traditional power grid scheduling fault disposal, can realize fast and accurate response strategy generation and execution in a high-dimensional complex power grid, and significantly improves the ability of the power grid to deal with sudden faults.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and more specifically, to a method for intelligent handling of power grid dispatching faults based on the collaborative operation of multiple agents. Background Technology

[0002] In modern society, the power system has become a critical infrastructure supporting social operation and development. Its safe and stable operation plays a vital role in ensuring the normal order of social and economic development and people's lives. With rapid economic development and the continuous improvement of people's living standards, the demand for electricity in the whole society continues to grow, placing higher demands on the reliability and stability of power supply. Once a power system failure occurs, it may trigger a chain reaction, leading to local or even large-scale power outages, causing huge losses to the social economy and severely impacting people's daily lives. For example, a large-scale power outage caused by a circuit breaker failure in a certain region paralyzed traffic and halted commercial activities. It is evident that ensuring the stable operation of the power system is of immeasurable importance for maintaining the normal operation of society and the sustainable development of the economy. As a key link in ensuring the safe and stable operation of the power system, grid dispatch fault handling aims to identify fault types and locate fault locations in a timely and accurate manner through real-time monitoring and analysis of the grid's operating status, and to provide corresponding fault handling strategies.

[0003] However, with the continuous expansion of the power system's scale and the increasing complexity of its structure, traditional power grid dispatching and fault handling face numerous challenges. On the one hand, the power grid contains numerous devices and a wide variety of fault types. The massive amounts of data generated when faults occur make it difficult for traditional methods to process and analyze them quickly and accurately, resulting in low handling efficiency. On the other hand, traditional handling methods often rely on a single data source or a fixed model, lacking the ability to effectively handle complex faults and uncertainties, and are difficult to adapt to the dynamic changes in the power grid's operating environment. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent method for handling power grid dispatching faults based on multi-agent collaborative operation. This method solves the weaknesses of traditional power grid dispatching fault handling and can achieve rapid and accurate response strategy generation and execution in high-dimensional complex power grids, significantly improving the power grid's ability to cope with sudden faults.

[0005] The technical solution of this invention is as follows:

[0006] Firstly, this application provides a method for intelligent handling of power grid dispatching faults based on multi-agent cooperative operation, which includes the following steps:

[0007] S1. Obtain the current operating status information of the power grid, and perform multi-dimensional data fusion on the current operating status information to construct the power grid state space;

[0008] S2. Calculate fault risk based on the results of multi-dimensional data fusion;

[0009] S3. Construct a multi-agent Markov decision process model using the failure risk calculation results;

[0010] S4. Based on the multi-agent Markov decision process model, an attention mechanism is introduced to perform dynamic feature aggregation;

[0011] S5. Calculate dynamic resource scheduling based on the dynamic feature aggregation results, and handle faults based on the scheduling results.

[0012] Furthermore, in step S1, the calculation formula for obtaining the current operating status information of the power grid includes:

[0013]

[0014]

[0015] In the formula, This represents the minimum sum of the products of the region's response time and the degree of fault impact on the power grid region. Let i be the response time of the i-th power grid region; The degree of fault impact in the power grid area; N is the total number of nodes in the power grid area. This represents the amount of electrical energy change between the i-th node and the j-th node. This represents the electrical energy between the i-th node and the j-th node at time t. Let be the electrical energy between the i-th node and the k-th node at time t; and All are adjustment coefficients; is the fault propagation coefficient of the k-th node; M is the number of nodes within the fault propagation range.

[0016] Furthermore, the calculation formula for calculating fault risk based on the results of multi-dimensional data fusion includes:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023] In the formula, For the i-th node in The probability of failure at any given time. Let be the failure probability of the i-th node at time t. Let be the failure risk of the i-th node at time t. The fault propagation coefficient is... Take an action in state s Expected value The immediate reward for the current decision, where s is the current state. For the next state, For the current action, For the next action, For state Take action below Maximum expected value, For electrical energy; As a comprehensive indicator, Let be the weight of the i-th node. This represents the degree of fault impact on the power grid area, with a value of 1 or 0, representing successful isolation or no isolation, respectively. The minimum value of the relevant indicators for the time required for all nodes to restore load, where N is the total number of nodes in the power grid area. The time required for the i-th node to restore its load. Let be the load of the i-th node at time t. For the node's maximum load, For state changes, For adjustment coefficients, Let be the influence coefficient between the i-th node and the j-th node. For the state change of the j-th node at time t, Let be the state of the i-th node at time t. For the load change of the i-th node, This represents the device state of the i-th node.

[0024] Furthermore, in step S3, the calculation formula for constructing the multi-agent Markov decision process model includes:

[0025]

[0026]

[0027]

[0028]

[0029]

[0030] In the formula, Let N represent the fault impact degree of power grid region i on region j, and N be the total number of nodes in the power grid region. For region i and region The influence coefficient between them Let k be the weight of the influence of region k on region j. i and j are respectively the first, second, third, fourth, fifth, sixth, seventh ... eighth, tenth, tenth There are nodes i and j. Let be the change in connectivity between the i-th node and the j-th node. This is the topological influence coefficient. Let be the connectivity between node k and node j at time t. For state Take action below Expected value Let the state at time t be... For the action selection at time t, The instant reward at time t, The fault propagation coefficient is... For the next action, Prioritize restoring load to region j for region i. Let be the load of the i-th region at time t. and All are adjustment coefficients. This refers to the change in the state of the power grid. Let be the fault propagation coefficient of the i-th node. Let represent the fault propagation effect of the i-th node at time t.

[0031] Further, step S4 includes:

[0032] An attention mechanism is introduced through a multi-agent Markov decision process model to highlight key relationships;

[0033] Based on feature vector combination and multilayer perceptron weight scoring, other nodes most relevant to the state changes of a specific agent are automatically identified.

[0034] By weighting and aggregating the state information of neighboring nodes using normalized attention weights, dynamic feature aggregation results are obtained.

[0035] Furthermore, step S4 also includes: introducing an asynchronous multi-channel experience playback communication protocol that supports priority weights through a multi-agent Markov decision process model to adapt to the power grid scenario.

[0036] Further, step S5 includes:

[0037] Analysis of historical fault data, power grid operation status, and regional load fluctuations is based on the aggregation results of dynamic features.

[0038] Based on the analysis results, a model with time series prediction capabilities is trained for prediction.

[0039] Based on the prediction results, instructions are issued to the edge computing nodes in the corresponding areas to pre-allocate computing resources, thereby achieving dynamic scheduling.

[0040] Fault handling is performed based on the scheduling results.

[0041] Secondly, this application provides an electronic device, comprising:

[0042] Memory, used to store one or more programs;

[0043] processor;

[0044] When one or more of the above programs are executed by the above processor, a method for intelligent handling of power grid dispatching faults based on multi-agent cooperative operation, as described in any of the first aspects above, is implemented.

[0045] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for intelligent handling of power grid dispatching faults based on multi-agent cooperative operation as described in any of the first aspects above.

[0046] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:

[0047] (1) The present invention provides a method for intelligent handling of power grid dispatching faults based on multi-agent collaborative operation. It integrates deep reinforcement learning and dynamic topology perception mechanism to construct a high-dimensional state space representation model for complex power grid environment. Through multi-agent Markov decision process modeling, it captures the heterogeneity and spatiotemporal correlation characteristics between devices, solves the weakness of traditional power grid dispatching fault handling, and can realize fast and accurate response strategy generation and execution in high-dimensional complex power grid, significantly improving the power grid's ability to cope with sudden faults;

[0048] (2) This invention designs a feature aggregation architecture based on dynamic graph attention network and introduces an asynchronous multi-channel experience playback communication protocol that supports priority weights to achieve efficient knowledge sharing of cross-regional intelligent agents;

[0049] (3) Based on the cloud-edge-device collaborative architecture, this invention constructs a distributed learning platform with dynamic resource scheduling and elastic scaling capabilities, and proposes a hierarchical sparse reward reshaping mechanism based on the spatiotemporal pyramid to accelerate strategy convergence and local intelligent evolution. Attached Figure Description

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart illustrating the steps of an intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation according to the present invention.

[0052] Figure 2 This is a schematic structural block diagram of an electronic device according to an embodiment of the present invention.

[0053] Icons: 101, memory; 102, processor; 103, communication interface. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0055] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0056] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0057] It should be noted that, in this document, the term "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0058] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the various embodiments and features described below can be combined with each other.

[0059] Example 1

[0060] Please see Figure 1 , Figure 1 The diagram shows the steps of a power grid dispatching fault intelligent handling method based on multi-agent cooperative operation provided in an embodiment of this application.

[0061] This invention provides an intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation, comprising the following steps:

[0062] S1. Obtain the current operating status information of the power grid, and perform multi-dimensional data fusion on the current operating status information to construct the power grid state space;

[0063] S2. Calculate fault risk based on the results of multi-dimensional data fusion;

[0064] S3. Construct a multi-agent Markov decision process model using the failure risk calculation results;

[0065] S4. Based on the multi-agent Markov decision process model, an attention mechanism is introduced to perform dynamic feature aggregation;

[0066] S5. Calculate dynamic resource scheduling based on the dynamic feature aggregation results, and handle faults based on the scheduling results.

[0067] Multi-dimensional data fusion includes integrating electrical parameters such as voltage, current, and frequency, as well as multi-source data such as node connection relationships, load changes, and equipment status.

[0068] In a preferred embodiment, the calculation formula for obtaining the current operating status information of the power grid in step S1 includes:

[0069]

[0070]

[0071] In the formula, This represents the minimum sum of the products of the region's response time and the degree of fault impact on the power grid region. Let i be the response time of the i-th power grid region; The degree of fault impact in the power grid area; N is the total number of nodes in the power grid area. This represents the amount of electrical energy change between the i-th node and the j-th node. This represents the electrical energy between the i-th node and the j-th node at time t. Let be the electrical energy between the i-th node and the k-th node at time t; and All are adjustment coefficients; is the fault propagation coefficient of the k-th node; M is the number of nodes within the fault propagation range.

[0072] As a preferred implementation method, the calculation formula for fault risk calculation based on multi-dimensional data fusion results includes:

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079] In the formula, For the i-th node in The probability of failure at any given time. Let be the failure probability of the i-th node at time t. Let be the failure risk of the i-th node at time t. The fault propagation coefficient is... Take an action in state s Expected value The immediate reward for the current decision, where s is the current state. For the next state, For the current action, For the next action, For state Take action below Maximum expected value, For electrical energy; As a comprehensive indicator, Let be the weight of the i-th node. This represents the degree of fault impact on the power grid area, with a value of 1 or 0, representing successful isolation or no isolation, respectively. The minimum value of the relevant indicators for the time required for all nodes to restore load, where N is the total number of nodes in the power grid area. The time required for the i-th node to restore its load. Let be the load of the i-th node at time t. For the node's maximum load, For state changes, For adjustment coefficients, Let be the influence coefficient between the i-th node and the j-th node. For the state change of the j-th node at time t, Let be the state of the i-th node at time t. For the load change of the i-th node, This represents the device state of the i-th node.

[0080] As a preferred implementation, in step S3, the calculation formula for constructing the multi-agent Markov decision process model includes:

[0081]

[0082]

[0083]

[0084]

[0085]

[0086] In the formula, Let N represent the fault impact degree of power grid region i on region j, and N be the total number of nodes in the power grid region. For region i and region The influence coefficient between them Let k be the weight of the influence of region k on region j. i and j are respectively the first, second, third, fourth, fifth, sixth, seventh ... eighth, tenth, tenth There are nodes i and j. Let be the change in connectivity between the i-th node and the j-th node. This is the topological influence coefficient. Let be the connectivity between node k and node j at time t. For state Take action below Expected value Let the state at time t be... For the action selection at time t, The instant reward at time t, The fault propagation coefficient is... For the next action, Prioritize restoring load to region j for region i. Let be the load of the i-th region at time t. and All are adjustment coefficients. This refers to the change in the state of the power grid. Let be the fault propagation coefficient of the i-th node. Let represent the fault propagation effect of the i-th node at time t.

[0087] In a preferred embodiment, step S4 includes:

[0088] An attention mechanism is introduced through a multi-agent Markov decision process model to highlight key relationships;

[0089] Based on feature vector combination and multilayer perceptron weight scoring, other nodes most relevant to the state changes of a specific agent are automatically identified.

[0090] By weighting and aggregating the state information of neighboring nodes using normalized attention weights, dynamic feature aggregation results are obtained.

[0091] It should be noted that by introducing a state correlation mechanism, and based on the combination of feature vectors between each pair of agents, a nonlinear structure such as a multilayer perceptron is used for weight scoring. This process can automatically identify other agent nodes that are most relevant to the state change of a certain agent within a specific time window. When the line current in a certain area fluctuates abnormally, the states of neighboring substations, switching equipment, and upstream and downstream load agents may also change synchronously. The dynamic attention graph network can assign higher attention weights to these nodes, thereby improving the sensitivity and accuracy of the overall fault handling strategy. The attention weights obtained after normalization are used to weighted aggregate the state information of neighboring nodes to obtain the high-level feature representation of the current node.

[0092] Therefore, by modeling the relationships between multiple agents through the structure of graph neural networks and introducing an attention mechanism to highlight key relationships, dynamic and accurate aggregation of agent features can be achieved.

[0093] As a preferred implementation, step S4 further includes: introducing an asynchronous multi-channel experience playback communication protocol that supports priority weights through a multi-agent Markov decision process model to adapt to the power grid scenario.

[0094] Therefore, by introducing an asynchronous multi-channel experience playback communication protocol that supports priority weighting, this approach adapts to the multiple challenges of high dynamism, heterogeneity, and limited communication resources in power grid scenarios. This promotes the rapid optimization and convergence of the intelligent fault handling system for power system dispatching, assigning priority labels to different experience samples to achieve priority scheduling and dissemination of critical experiences. Regarding the transmission mechanism, an asynchronous multi-channel structure is introduced to overcome the bandwidth bottleneck of traditional centralized communication. The multi-channel structure means that the communication system opens multiple data channels in parallel, each carrying experience samples of different priorities, categories, or types. Each channel adopts adaptive coding and transmission strategies based on the attributes of the experience samples and the communication resource status, ensuring that important information can still be stably and efficiently transmitted to the target agent even under conditions of network congestion, channel interference, or node asynchrony. In power grid edge areas or nodes with weak communication capabilities, the multi-channel strategy significantly improves system robustness and real-time response capabilities. The introduction of the asynchronous transmission mechanism solves problems such as uneven distribution of multiple agents and inconsistent communication delays in the power grid environment. Different agents can send or receive experience data independently at their respective time steps without waiting for global synchronization, thus avoiding experience loss or update delays caused by time lag.

[0095] In a preferred embodiment, step S5 includes:

[0096] Analysis of historical fault data, power grid operation status, and regional load fluctuations is based on the aggregation results of dynamic features.

[0097] Based on the analysis results, a model with time series prediction capabilities is trained for prediction.

[0098] Based on the prediction results, instructions are issued to the edge computing nodes in the corresponding areas to pre-allocate computing resources, thereby achieving dynamic scheduling.

[0099] Fault handling is performed based on the scheduling results.

[0100] It's important to note that, to effectively respond to sudden computational pressures, the power system has introduced a dynamic scheduling mechanism based on load forecasting. This mechanism analyzes multi-source heterogeneous data, including historical fault data, grid operating status, and regional load fluctuations, to train a model with time-series forecasting capabilities. This model predicts and warns of potential high-load scenarios in the near future. During seasonal changes or peak electricity consumption periods in the power system, the load model can identify potential large-scale load changes in advance and issue instructions to pre-allocate computing resources to edge computing nodes in the corresponding areas. This allows for necessary preparations before the actual load arrives, avoiding system performance bottlenecks or delays caused by sudden computing demands. As shown in the spatiotemporal assessment map of the area affected by grid faults, when the system predicts a surge in load or actually detects a backlog of computing tasks or task response timeouts, the cloud platform can quickly achieve elastic scaling based on containerization technology. Kubernetes, a container orchestration technology, can automatically schedule and deploy new computing containers within seconds, dynamically increasing the parallelism of training tasks or the number of model inference threads to improve system response speed.

[0101] Example 2

[0102] Please see Figure 2 , Figure 2 This is a schematic structural block diagram of an electronic device provided in an embodiment of this application.

[0103] An electronic device includes a memory 101, a processor 102, and a communication interface 103. The memory 101, processor 102, and communication interface 103 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines. The memory 101 can be used to store software programs and modules. The processor 102 executes the software programs and modules stored in the memory 101 to perform various functional applications and data processing. The communication interface 103 can be used for signaling or data communication with other node devices.

[0104] The memory 101 may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), etc.

[0105] The processor 102 can be an integrated circuit chip with signal processing capabilities. The processor 102 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0106] It is understood that the structure shown in the figure is for illustrative purposes only. A method for intelligent handling of power grid dispatching faults based on multi-agent collaborative operation may include more or fewer components than shown in the figure, or have a different configuration. The components shown in the figure can be implemented in hardware, software, or a combination thereof.

[0107] In the embodiments provided in this application, it should be understood that the disclosed methods can also be implemented in other ways. The embodiments described above are merely illustrative. For example, the flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this application. In this regard, each block in the flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0108] In addition, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0109] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0110] In summary, this invention integrates deep reinforcement learning and dynamic topology sensing mechanisms to construct a high-dimensional state-space representation model for complex power grid environments. Through multi-agent Markov decision process modeling, it captures the heterogeneity and spatiotemporal correlation characteristics among devices. At the agent collaboration level, a feature aggregation architecture based on a dynamic graph attention network is designed, and an asynchronous multi-channel experience playback communication protocol supporting priority weights is introduced to achieve efficient knowledge sharing among agents across regions. At the system deployment level, based on a cloud-edge-device collaborative architecture, a distributed learning platform with dynamic resource scheduling and elastic scaling capabilities is constructed. A hierarchical sparse reward reshaping mechanism based on a spatiotemporal pyramid is proposed to accelerate policy convergence and local intelligent evolution. Reinforcement learning demonstrates powerful problem-solving capabilities in fields such as automatic control, intelligent game theory, and robot navigation.

[0111] Through continuous interaction between intelligent agents and the environment, optimal behavioral strategies are learned from feedback, making it suitable for dynamic, complex, and unpredictable system scenarios. Further development of multi-agent reinforcement learning technology has broken through the limitations of single-agent systems, enabling the simulation of collaboration, competition, and game-theoretic processes among multiple agents, better reflecting the actual situation of multiple subsystems or nodes working together in real-world systems. In intelligent fault handling systems for power grid dispatching, each region, each piece of equipment, and even each control node can be considered an intelligent agent. Their collaborative efforts after a fault occurs directly determine the speed and quality of fault control and system recovery. Multi-agent reinforcement learning technology can precisely simulate and optimize this complex collaborative relationship, providing solid technical support for building a collaborative and efficient intelligent emergency response system. The core objective of applying multi-agent reinforcement learning to the optimization of intelligent fault handling systems for power grids is to build a comprehensive platform with high perception capabilities, autonomous learning capabilities, and intelligent decision-making capabilities. Each agent in the system can make preliminary response decisions based on local observation information and achieve globally optimal collaborative control through information sharing and interaction with other agents.

[0112] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0113] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from the spirit or essential characteristics of this application. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of this application is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within this application. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for intelligent handling of power grid dispatching faults based on multi-agent cooperative operation, characterized in that, Includes the following steps: S1. Obtain the current operating status information of the power grid, and perform multi-dimensional data fusion on the current operating status information to construct the power grid state space; S2. Calculate fault risk based on the results of multi-dimensional data fusion; S3. Construct a multi-agent Markov decision process model using the failure risk calculation results; S4. Based on the multi-agent Markov decision process model, an attention mechanism is introduced to perform dynamic feature aggregation; S5. Calculate dynamic resource scheduling based on the dynamic feature aggregation results, and handle faults based on the scheduling results.

2. The intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation as described in claim 1, characterized in that, In step S1, the calculation formula for obtaining the current operating status information of the power grid includes: ， ， In the formula, This represents the minimum sum of the products of the region's response time and the degree of fault impact on the power grid region. Let i be the response time of the i-th power grid region; The degree of fault impact in the power grid area; N is the total number of nodes in the power grid area. This represents the amount of electrical energy change between the i-th node and the j-th node. This represents the electrical energy between the i-th node and the j-th node at time t. Let be the electrical energy between the i-th node and the k-th node at time t; and All are adjustment coefficients; is the fault propagation coefficient of the k-th node; M is the number of nodes within the fault propagation range.

3. The intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation as described in claim 1, characterized in that, The calculation formula for fault risk calculation based on multi-dimensional data fusion results. include: ， ， ， ， ， ， In the formula, For the i-th node in The probability of failure at any given time. Let be the failure probability of the i-th node at time t. Let be the failure risk of the i-th node at time t. The fault propagation coefficient is... Take an action in state s Expected value The immediate reward for the current decision, where s is the current state. For the next state, For the current action, For the next action, For state Take action below Maximum expected value, For electrical energy; As a comprehensive indicator, Let be the weight of the i-th node. This represents the degree of fault impact on the power grid area, with a value of 1 or 0, representing successful isolation or no isolation, respectively. The minimum value of the relevant indicators for the time required for all nodes to restore load, where N is the total number of nodes in the power grid area. The time required for the i-th node to restore its load. Let be the load of the i-th node at time t. For the node's maximum load, For state changes, For adjustment coefficients, Let be the influence coefficient between the i-th node and the j-th node. For the state change of the j-th node at time t, Let be the state of the i-th node at time t. For the load change of the i-th node, This represents the device state of the i-th node.

4. The intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation as described in claim 1, characterized in that, In step S3, the calculation formula for constructing the multi-agent Markov decision process model includes: ， ， ， ， ， In the formula, Let N represent the fault impact degree of power grid region i on region j, and N be the total number of nodes in the power grid region. For region i and region The influence coefficient between them Let k be the weight of the influence of region k on region j. i and j are respectively the first, second, third, fourth, fifth, sixth, seventh ... eighth, tenth, tenth There are nodes i and j. Let be the change in connectivity between the i-th node and the j-th node. This is the topological influence coefficient. Let be the connectivity between node k and node j at time t. For state Take action below Expected value Let the state at time t be... For the action selection at time t, The instant reward at time t, The fault propagation coefficient is... For the next action, Prioritize restoring load to region j for region i. Let be the load of the i-th region at time t. and All are adjustment coefficients. This refers to the change in the state of the power grid. Let be the fault propagation coefficient of the i-th node. Let represent the fault propagation effect of the i-th node at time t.

5. The intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation as described in claim 1, characterized in that, Step S4 includes: An attention mechanism is introduced through a multi-agent Markov decision process model to highlight key relationships; Based on feature vector combination and multilayer perceptron weight scoring, other nodes most relevant to the state changes of a specific agent are automatically identified. By weighting and aggregating the state information of neighboring nodes using normalized attention weights, dynamic feature aggregation results are obtained.

6. The intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation as described in claim 5, characterized in that, Step S4 also includes: introducing an asynchronous multi-channel experience playback communication protocol that supports priority weights through a multi-agent Markov decision process model to adapt to the power grid scenario.

7. The intelligent fault handling method for power grid dispatching based on multi-agent cooperative operation as described in claim 1, characterized in that, Step S5 includes: Analysis of historical fault data, power grid operation status, and regional load fluctuations is based on the aggregation results of dynamic features. Based on the analysis results, a model with time series prediction capabilities is trained for prediction. Based on the prediction results, instructions are issued to the edge computing nodes in the corresponding areas to pre-allocate computing resources, thereby achieving dynamic scheduling. Fault handling is performed based on the scheduling results.

8. An electronic device, characterized in that, include: Memory, used to store one or more programs; processor; When the processor executes the one or more programs, it implements a method for intelligent handling of power grid dispatching faults based on multi-agent cooperative operation as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a method for intelligent handling of power grid dispatching faults based on multi-agent cooperative operation as described in any one of claims 1-7.

Images