On-line calculation system and method for stability boundary of tidal flow section based on multiple agents

Abstract

The invention provides a power flow section stability boundary online calculation system and method based on multiple agents. The method comprises the steps that power grid operation state data, a power grid power flow section, the increasing direction of a power vector of the power grid power flow section and a voltage phase angle initial value of a boundary bus are obtained; power grid regions are divided based on the power grid operation state data, and each region is correspondingly provided with a region intelligent agent; determining a to-be-analyzed power grid power flow section in the power grid power flow sections by using the central coordination agent, packaging the increasing direction of the power vector of the to-be-analyzed power grid power flow section and the initial value of the voltage phase angle of the boundary bus into a calculation instruction, and issuing the calculation instruction to the regional agent; adopting an MAPPO algorithm to calculate the stability boundary of the to-be-analyzed power grid power flow section of the region; and removing abnormal data, and generating an analyzed power grid power flow section stability boundary. According to the invention, the stability and reliability of multi-agent cooperation can be guaranteed, and the intellectualization and high efficiency of power grid power flow section stability boundary calculation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial intelligence, in particular to a multi-agent-based online calculation system and method for stability boundaries of power flow sections. BACKGROUND

[0002] With the promotion of the global "double carbon" goal and the acceleration of energy structure transformation, the penetration rate of new energy such as wind power and photovoltaic continues to rise. However, the traditional calculation method has a long calculation period and cannot respond to real-time changes in power grid topology, resulting in response lag, poor adaptability, insufficient collaboration and other problems, and cannot meet the millisecond-level online scheduling requirements. In the prior art, the single-agent method is limited by the computing power and cannot handle large power grid multi-dimensional state data, and simple partition regulation is prone to boundary calculation deviation due to information fragmentation between regions.

[0003] In view of the above problems, there is an urgent need for a multi-agent-based online calculation system and method for stability boundaries of power flow sections. SUMMARY

[0004] The present application aims to provide a multi-agent-based online calculation system and method for stability boundaries of power flow sections to solve the technical problems raised in the background.

[0005] To achieve the above-mentioned purpose, the first aspect of the present application proposes a multi-agent-based online calculation system for stability boundaries of power flow sections, which comprises: An acquisition module is configured to obtain power grid operating state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters, power flow sections of the power grid, and power vector growth direction of the power flow sections, initial voltage phase angle value of boundary buses and power fluctuation data of regional tie lines; A partition module is configured to divide the power grid region based on the power grid operating state data using a clustering analysis method, wherein each partition is configured with a regional agent; An initialization module is configured to determine a power flow section to be analyzed in the power flow section of the power grid by using a central coordination agent according to the scheduling requirement and the power fluctuation data of the regional tie line combination; An interaction module is configured to encapsulate the power vector growth direction of the power flow section to be analyzed and the initial voltage phase angle value of the boundary bus as a calculation instruction by using the central coordination agent, and synchronously issue the calculation instruction to the regional agent where the power flow section to be analyzed is located; A boundary calculation module is configured to calculate the stability boundary of the power flow section to be analyzed in the region by using the regional agent according to the calculation instruction received from the central coordination agent; The aggregation module is configured to check the validity of the stability boundary of the power flow section uploaded by all regional agents by using the central coordination agent, eliminate abnormal data, and generate the stability boundary of the analyzed power flow section.

[0006] Further, the specific method for dividing the power grid region by using the clustering analysis method comprises: standardizing the power grid operation state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters to obtain standardized power grid operation data; performing clustering analysis on the standardized power grid operation data by using a clustering algorithm, and dividing the power grid into multiple regions with similar electrical characteristics according to the clustering results; wherein the nodes in each region have similar voltage levels and power distribution.

[0007] Further, the specific method for determining the power flow section to be analyzed in the power flow section according to the dispatching demand and the power grid operation state data comprises: determining the load sensitivity of the regional tie-line power fluctuation data in the obtained power flow section to the dispatching demand by a sensitivity calculation application; selecting the power flow section corresponding to the top three regional tie-lines with high load sensitivity as the power flow section to be analyzed.

[0008] Further, the specific method for calculating the stability boundary of the power flow section to be analyzed in the region by using the MAPPO algorithm comprises: The MAPPO algorithm drives policy update by loss function optimization, limits the update amplitude of new and old policies, and ensures that the policy is optimized in the direction of reducing boundary calculation error; updating the new and old policies by an update mechanism; iterating the experience collection and loss function optimization process until the objective function converges, and finally outputting the stability boundary result.

[0009] Further, the objective function of the policy update of the MAPPO algorithm is defined as follows: wherein, is the ratio of the new policy to the old policy in the region, and the calculation formula is: ; B is the size of the training batch; n is the number of related operations of each agent in each batch; is the estimated value of the advantage function, which is used to measure the advantage of the action in the state ; is the entropy of the policy, is a hyperparameter used to control the size of the policy update step. The model is trained, wherein The local observation information of the agent .

[0010] Further, the loss function of the MAPPO algorithm It is defined as follows: Where B is the training batch size; n is the number of related operations of each agent in each batch; is the estimated value of the advantage function, which is used to measure the advantage of the action In the state ; is the entropy of the policy, is a hyperparameter used to control the size of the policy update step; is the estimated value of the real reward by the current evaluation network, is the expected value of the round reward of the sampled data, is the estimated value of the real reward by the old evaluation network.

[0011] Further, the stability boundary effectiveness of the power grid flow section uploaded by all regional agents is checked, and abnormal data is removed, and the specific method for generating the stability boundary of the analyzed power grid flow section is as follows: The central agent removes the stability boundary with power greater than the preset value uploaded by the regional agent, and obtains the effective stability boundary of the regional agent; The stability boundaries of all effective regional agents are spliced to generate the generated stability boundary of the analyzed power grid flow section.

[0012] Further, it further comprises: An output module is used for the central coordinating agent to output the stability boundary of all flow sections to other power grid dispatching systems, and to provide basis for dispatching decision.

[0013] The second aspect of the present application proposes a multi-agent based online calculation method for the stability boundary of the flow section, comprising the following steps: Obtain the power grid operating state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters, power grid flow section, and the growth direction of the power vector of the power grid flow section and the initial value of the voltage phase angle of the boundary bus; Based on the power grid operating state data, the power grid region is divided by using the clustering analysis method, wherein each partition corresponds to a regional agent; The central coordinating agent determines the power grid flow section to be analyzed in the power grid flow section according to the power fluctuation data of the dispatching demand and the regional tie line combination; The central coordination agent encapsulates the growth direction of the power vector of the power flow section to be analyzed and the initial value of the phase angle of the voltage of the boundary bus as a calculation instruction, and synchronously issues the calculation instruction to the regional agent of the area where the power flow section to be analyzed is located. The regional agent calculates the stable boundary of the power flow section to be analyzed in the region according to the calculation instruction received from the central coordination agent by using the MAPPO algorithm. The central coordination agent checks the validity of the stable boundary of the power flow section uploaded by all regional agents, eliminates abnormal data, and generates the stable boundary of the analyzed power flow section.

[0014] The third aspect of the present application provides a computer program product comprising computer programs / instructions, which, when executed by a processor, implement the steps of the multi-agent-based power flow section stable boundary online calculation method.

[0015] The beneficial effects brought by the technical scheme of the present application include: taking advantage of the synergy of the multi-agent system, the central coordination agent and the regional agent have clear division of labor and efficient interaction, which not only ensures the overall stability of the boundary calculation, but also takes advantage of the accurate perception and rapid response ability of the regional agent to the local power grid state, effectively improves the safety margin of the power grid, reduces the risk of power flow exceeding limit caused by new energy fluctuation or cross-regional transmission change, and ensures the safety and stability of the power grid operation. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The system schematic diagram of an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to enable personnel in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0018] First, some technical terms in the present application are explained and described in order to facilitate the understanding of the present application by those skilled in the art.

[0019] The regional agent is an artificial intelligence system focusing on a specific geographical space, which can integrate regional data, make autonomous decisions and perform tasks. The core is to enable the specific region to have the comprehensive ability of "perception, thinking and action". The core of the central coordination agent is "globality and coordination", which does not directly handle specific matters of a single region, but is responsible for integrating data of multiple regions, balancing global resources and solving cross-regional problems.

[0020] Power flow calculation is a basic calculation in power system analysis, which is used to study the steady-state operation of the power system. According to the given power grid structure, parameters and operating conditions of generator, load and other elements, the operating state of each part of the power system is determined, such as the voltage amplitude and phase angle of each bus, the power distribution of each branch, the power loss of the network, etc.

[0021] The purpose of power flow calculation is to check whether each element of the power system is overloaded, and whether the voltage of each bus is within the allowed range, to provide basis for power system planning, operation mode determination, relay protection setting, safety analysis, etc.

[0022] In order to make the purpose, technical scheme and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0023] Embodiment 1 The first aspect of the present application proposes a multi-agent based power flow section stability boundary online calculation system, as shown in the figure, the system comprises: Figure 1 The acquisition module is used to acquire power grid operating state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters, power grid power flow section and the growth direction of power vector of the power grid power flow section, the initial value of voltage phase angle of boundary bus and the power fluctuation data of regional tie line combination. Specifically, the 220kV and above voltage level power grid operating state data is collected from the energy management system or the data acquisition and monitoring control system, including: Bus parameters: node voltage level, name, type, such as PQ, PV, balanced node, etc.

[0024] Transmission line parameters: resistance, reactance, conductance, susceptance, thermal stability limit, length.

[0025] Transformer parameters: transformation ratio, impedance, tap range.

[0026] Generator parameters: rated power, maximum / minimum output, excitation system parameters, prime mover model parameters, speed regulation system parameters.

[0027] Load parameters: active power, reactive power, load model, such as constant impedance, constant power, etc.

[0028] The EMS dispatching automation system is used to acquire the power grid power flow section and the growth direction of power vector of the power grid power flow section, the initial value of voltage phase angle of boundary bus and the power fluctuation data of regional tie line combination. ​The partition module is configured to divide the power grid region by using a clustering analysis method based on power grid operation state data, wherein each partition corresponds to a regional intelligent agent; Further, the specific method for dividing the power grid region by using the clustering analysis method includes: The power grid operation state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters are standardized to obtain standardized power grid operation data; The clustering algorithm is used to analyze the standardized power grid operation data, and the power grid is divided into multiple regions with similar electrical characteristics according to the clustering results; Wherein, the nodes in each region have similar voltage level and power distribution.

[0029] Specifically, the power grid operation state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters are standardized to obtain standardized power grid operation data; the standardized power grid operation data is input to the power grid region division application; according to the input power grid operation state data, the power flow calculation software application module is called through the interface to perform power flow calculation, and the sensitivity matrix S is obtained; based on the sensitivity matrix S, the electrical distance matrix D is calculated by using the Euclidean distance formula; the data of the electrical distance matrix D is normalized to obtain a fuzzy matrix; the dynamic clustering graph is obtained by using the transitive closure method for initial partitioning, which shows the devices classified into the same class, and the same class is divided into a region; remove the isolated nodes and integrate them into the region with the strongest electrical coupling; nodes connected to multiple PQs are merged into PV nodes; check the power balance condition to obtain the partitioning result.

[0030] The initialization module is configured to determine the power grid flow section to be analyzed in the power grid flow section by using the central coordination intelligent agent according to the power fluctuation data of the regional tie line combination and the scheduling demand; Further, the specific method for determining the power grid flow section to be analyzed in the power grid flow section according to the scheduling demand and the power grid operation state data includes: The power fluctuation data of the regional tie line in the obtained power grid flow section is determined to have a load sensitivity to the scheduling demand by using the sensitivity calculation application; The power grid flow section corresponding to the top three regional tie lines with the highest load sensitivity is selected as the power grid flow section to be analyzed.

[0031] Specifically, the dispatching demand focus may affect the weak link of global stability. For example, during the peak summer, the power supply of load centers such as urban core areas and industrial parks needs to be prioritized, and the new energy consumption problem needs to be considered due to the large output of wind and photovoltaic power. For important user power supply requirements, such as hospitals, data centers, and railway hubs, the power supply path needs to be reliable. The equipment that has already failed and has already generated risks needs to be eliminated.

[0032] By calling the sensitivity calculation application, the influence degree of the power change of each tie line on these links is obtained, and the power flow section of the regional tie line combination that is prone to cause safety problems is selected as the power flow section to be analyzed.

[0033] The interaction module is configured to encapsulate the growth direction of the power vector of the power flow section to be analyzed and the initial value of the voltage phase angle of the boundary bus as a calculation instruction, and synchronously issue the calculation instruction to the regional agent where the power flow section to be analyzed is located. Specifically, encapsulating the growth direction of the power vector of the power flow section to be analyzed and the initial value of the voltage phase angle of the boundary bus as a calculation instruction includes: converting the growth direction of the power vector of the power flow section to be analyzed and the initial value of the voltage phase angle of the boundary bus into a computer-recognizable instruction in JSON format, and encapsulating. In terms of protocol adaptation, the regional agent adopts standardized protocols such as TCP / IP and MQTT to ensure the compatibility of cross-agent data interaction and avoid instruction loss or parsing errors; in terms of instruction issuance, the central coordination agent synchronously issues the calculation instruction to all regional agents through a broadcast mechanism to ensure the timeliness of parallel computing of each region and prevent the computing rhythm from being out of sync due to time difference in receiving instructions; in terms of data reception, the regional agent receives the preliminary calculation results of the regional stability boundary uploaded by the regional agent.

[0034] The boundary calculation module is configured to calculate the stability boundary of the power flow section to be analyzed in the region by using the regional agent according to the calculation instruction received from the central coordination agent. Specifically, the growth direction of the power vector of the power flow section in the calculation instruction specifies the direction in which the power of the section grows to the safety limit threshold; the bus voltage phase angle itself is a key parameter affecting the line flow, and is an important input parameter of the MAPPO model.

[0035] Further, the specific method for calculating the stability boundary of the power flow section to be analyzed in the region by using the MAPPO algorithm is as follows: The MAPPO algorithm drives policy update through loss function optimization, limits the update amplitude of new and old policies, and ensures that the policy is optimized in the direction of reducing boundary calculation error; The new and old policies are updated through an updating mechanism; Through repeated iteration of experience collection and loss function optimization process, until the objective function converges, finally output stable boundary result.

[0036] Specifically, the MAPPO algorithm process is: First, orthogonal initialization is performed to determine the policy network parameters and the value evaluation network parameters . At the same time, the learning rate α, the clipping coefficient , the entropy hyperparameter , the experience pool capacity, the data block length L and other key hyperparameters are set.

[0037] Experience is collected in small batches: each small batch is initialized as an empty trajectory , and each time step t in the task period is traversed. Each agent generates an action probability distribution according to the current local observation and the policy network , and samples an action ; the value evaluation network outputs a value estimate combined with the hidden layer state and the global state . After the agent performs the action, the next observation, state and reward are obtained, and the hidden layer state is updated. These data are stored in the trajectory . After the small batch task is completed, the trajectory is divided into data blocks according to the length and stored in the experience pool D.

[0038] For each trajectory , the generalized advantage estimation method is used to calculate the advantage function through multi-step temporal difference error, which improves the accuracy of advantage estimation; at the same time, the cumulative discounted reward is calculated based on the immediate reward and the discount factor, which is used as the supervision signal of the value evaluation network. Finally, the PopArt normalization technique is used to standardize and , eliminating the interference of data dimension difference on training.

[0039] Data blocks are extracted from the experience pool D in mini batches, and the first hidden layer state of the sub-data block is used to update the network RNN hidden layer state. When optimizing the policy network, the probability ratio of the new and old policies is calculated, which is substituted into the objective function containing the function and the entropy term, and the parameters are updated through the Adam optimizer; when optimizing the value evaluation network, the original error and are calculated.Loss of error , the parameters are updated with Adam optimizer , the process is repeated until the network loss is satisfactory or the number of iterations is exhausted.

[0040] After training, each agent disconnects from the central controller and only relies on its own real-time local observation and the trained policy network independently generate optimal actions to achieve distributed execution of multi-agent collaborative tasks.

[0041] Further, the target function of policy update of MAPPO algorithm is defined as follows: Where, is the ratio of the new policy to the old policy in the region, and the calculation formula is: ; B is the training batch size; n is the number of related operations of each agent in each batch; is the estimated value of the advantage function, which measures the advantage of the action in state ; is the entropy of the policy, is a hyperparameter used to control the size of the policy update step; Train the model, where is the local observation information of the agent .

[0042] Further, the loss function of the MAPPO algorithm is defined as follows: Where, B is the training batch size; n is the number of related operations of each agent in each batch; is the estimated value of the advantage function, which measures the advantage of the action in state ; is the entropy of the policy, is a hyperparameter used to control the size of the policy update step; is the current evaluation network's estimate of the true reward, is the expected value of the episode reward of the sampled data, is the old evaluation network's estimate of the true reward.

[0043] The summary module is used to check the stability boundary effectiveness of the power flow section uploaded by all regional agents using the central coordination agent, eliminate abnormal data, and generate the analyzed stability boundary of the power flow section.

[0044] Further, the central coordination agent checks the validity of the stability boundary of the power flow section uploaded by all regional agents, eliminates abnormal data, and generates the specific method for generating the stability boundary of the analyzed power flow section, which comprises: The central coordination agent eliminates the stability boundary uploaded by the regional agent with power greater than the preset value, and obtains the effective stability boundary of the regional agent. The stability boundaries of all effective regional agents are spliced to generate the stability boundary of all power flow sections.

[0045] Further, it further comprises: An output module for the central coordination agent to output the stability boundary of all power flow sections to other power grid dispatching systems to provide basis for dispatching decision.

[0046] Embodiment 2 The second aspect of the present application proposes an online calculation method for the stability boundary of the power flow section based on multiple agents, which comprises the following steps: Obtain the power grid operating state data including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters, power flow section and power vector growth direction of the power flow section, initial value of voltage phase angle of boundary bus and power fluctuation data of regional tie line combination; Based on the power grid operating state data, the power grid region is divided by using the clustering analysis method, wherein each partition corresponds to a regional agent; The central coordination agent determines the power flow section to be analyzed in the power flow section according to the dispatching demand and the power fluctuation data of the regional tie line combination; The central coordination agent encapsulates the power vector growth direction of the power flow section to be analyzed and the initial value of the voltage phase angle of the boundary bus as a calculation instruction, and synchronously issues the calculation instruction to the regional agent where the power flow section to be analyzed is located; The regional agent calculates the stability boundary of the power flow section to be analyzed in the region according to the received calculation instruction issued from the central coordination agent by using the MAPPO algorithm; The central coordination agent checks the validity of the stability boundary of the power flow section uploaded by all regional agents, eliminates abnormal data, and generates the stability boundary of the analyzed power flow section.

[0047] Embodiment 3 The third aspect of the present application proposes a computer program product comprising computer programs / instructions, which are executed by a processor to realize the steps of embodiment 2.

[0048] Those aspects of the specification that are not otherwise fully described are deemed to be part of the prior art. Those skilled in the art will appreciate that embodiments of the present application can be practiced in a method, system, or computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, magnetic disks, CD-ROMs, optical storage media such as DVD s, etc.) embodying computer readable program code.

[0049] The present application is described in reference to the flow diagrams and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flow diagrams and / or block diagrams, and combinations of blocks in the flow diagrams and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flow diagrams and / or block diagrams block or blocks. Figure 1 one or more functions specified in one or more of the flow diagram or block diagram blocks. Figure 1 one or more functions specified in one or more of the flow diagram or block diagram blocks.

[0050] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flow diagram or block diagram block or blocks. Figure 1 one or more functions specified in one or more of the flow diagram or block diagram blocks. Figure 1 one or more functions specified in one or more of the flow diagram or block diagram blocks.

[0051] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flow diagram or block diagram block or blocks. Figure 1 one or more functions specified in one or more of the flow diagram or block diagram blocks. Figure 1 one or more functions specified in one or more of the flow diagram or block diagram blocks.

[0052] It should be pointed out finally that the above examples are only used for illustrating the technical solutions of the present application but not for limiting the protection scope thereof, and although the present application has been described in detail with reference to the above examples, it should be understood by those skilled in the art that the specific embodiments of the present application can be changed, modified or replaced equivalently by those skilled in the art after reading the present application, but these changes, modifications or equivalent replacements are all within the protection scope of the claims to be filed of the present application.

Claims

1. An online calculation system for the stability boundary of power flow sections based on multi-agent systems, characterized in that, The system includes: The acquisition module is used to acquire power grid operating status data, including bus parameters, transmission line parameters, transformer parameters, generator parameters and load parameters, power grid power flow section and the growth direction of the power vector of the power flow section, the initial value of the voltage phase angle of the boundary bus and the power fluctuation data of the regional tie line combination; The partitioning module is used to divide the power grid into regions based on power grid operation status data using cluster analysis. Each partition is configured with a corresponding regional agent. The initialization module is used to determine the power flow section to be analyzed in the power flow section of the power grid based on the power fluctuation data of the central coordinating agent according to the scheduling requirements and the combination of regional tie lines. The interaction module is used to encapsulate the growth direction of the power vector of the power flow section of the power grid to be analyzed and the initial value of the voltage phase angle of the boundary bus into calculation instructions using the central coordinating agent, and synchronously send the calculation instructions to the regional agent where the power flow section of the power grid to be analyzed is located. The boundary calculation module is used to calculate the stability boundary of the power flow section of the power grid to be analyzed in the region by using the MAPPO algorithm based on the calculation instructions received from the central coordinating agent. The aggregation module is used to verify the validity of the stability boundaries of the power flow sections uploaded by all regional agents using the central coordinating agent, remove abnormal data, and generate the analyzed stability boundaries of the power flow sections.

2. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 1, characterized in that, Specific methods for dividing power grid areas using cluster analysis include: Standardized power grid operation data, including bus parameters, transmission line parameters, transformer parameters, generator parameters, and load parameters, are processed to obtain standardized power grid operation data. Clustering algorithms are used to perform cluster analysis on the standardized power grid operation data, and the power grid is divided into multiple regions with similar electrical characteristics based on the clustering results; Within each region, the nodes have similar voltage levels and power distributions.

3. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 1, characterized in that, The specific methods for determining the power flow section to be analyzed in the power flow section of the power grid based on dispatch demand and power fluctuation data of regional tie line combinations include: By applying sensitivity calculations, the load sensitivity of regional tie-line power fluctuation data in the acquired power flow sections of the power grid to dispatch demand is determined. The power flow sections corresponding to the three regional tie lines with the highest load sensitivity are selected as the power flow sections to be analyzed.

4. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 1, characterized in that, The specific method for calculating the stability boundary of the power flow section of the power grid to be analyzed in this region using the MAPPO algorithm is as follows: The MAPPO algorithm drives policy updates through loss function optimization, limits the update magnitude of the old and new policies, and ensures that the policy is optimized in the direction of reducing boundary computation errors. Update old and new strategies through an update mechanism; After repeated iterations of experience collection and loss function optimization, until the objective function converges, the final stable boundary result is output.

5. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 4, characterized in that, Objective function of policy update in MAPPO algorithm Defined as follows: in, The ratio of the new strategy to the old strategy in this region is calculated using the following formula: B is the training batch size; n is the number of operations performed by each agent within each batch. It is an estimate of the advantage function, used to measure action. In state The advantages of the following; For the entropy of the strategy, It is a hyperparameter used to control the size of the policy update step; The model is trained, where For intelligent agents Local observation information.

6. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 4, characterized in that, Loss function of MAPPO algorithm Defined as follows: Where B is the training batch size; n is the number of operations performed by each agent within each batch; It is an estimate of the advantage function, used to measure action. In state The advantages of the following; For the entropy of the strategy, It is a hyperparameter used to control the size of the policy update step; This is the current evaluation network's estimate of the true reward. The expected value of the round reward for the sampled data. It is the old evaluation network's estimate of the actual reward.

7. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 1, characterized in that, The specific method for verifying the validity of the stability boundaries of the power flow sections uploaded by all regional agents, removing outlier data, and generating the analyzed stability boundaries of the power flow sections includes: The central agent removes stable boundaries whose power exceeds a preset value from the regional agents, thus obtaining effective stable boundaries for the regional agents. The stable boundaries of all valid regional agents are spliced ​​together to generate the stable boundaries of the power flow section of the generated and analyzed power grid.

8. The online calculation system for the stability boundary of power flow section based on multi-agent as described in claim 1, characterized in that, Also includes: Output module: Used by the central coordinating agent to output the stability boundaries of all power flow sections to other power grid dispatching systems, providing a basis for dispatching decisions.

9. A method for online calculation of the stability boundary of power flow section based on multi-agent systems, comprising the following steps: Acquire power grid operating status data, including bus parameters, transmission line parameters, transformer parameters, generator parameters, and load parameters; power grid flow section; the growth direction of the power vector of the power flow section; and the initial value of the voltage phase angle of the boundary bus. Based on power grid operation status data, a cluster analysis method is used to divide the power grid into regions, with each region corresponding to a regional intelligent agent. The central coordinating agent uses power fluctuation data based on scheduling needs and regional tie-line combinations to determine the power flow sections to be analyzed in the power flow sections. The central coordinating agent encapsulates the growth direction of the power vector of the power flow section to be analyzed and the initial value of the voltage phase angle of the boundary bus into calculation instructions, and synchronously sends the calculation instructions to the regional agent where the power flow section of the power flow to be analyzed is located. Based on the calculation instructions received from the central coordinating agent, the regional agent uses the MAPPO algorithm to calculate the stability boundary of the power flow section of the power grid to be analyzed in the region. The central coordinating agent is used to verify the validity of the stability boundaries of the power flow sections uploaded by all regional agents, and abnormal data is removed to generate the stability boundaries of the analyzed power flow sections.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method of claim 9.

Images