Large power grid dispatching method and intelligent agent model adapted to power digital simulation system

Abstract

The present application discloses a large power grid dispatching method and intelligent agent model adapted to an electric power digital simulation system, belonging to the technical field of electric power dispatching. The method comprises: inputting the power grid topology data of the electric power system into an intelligent prediction model, making decisions on preset problems, and obtaining decision tasks; the intelligent prediction model is trained by historical simulation data; the historical simulation data is simulated by an electric power digital simulation system; the decision tasks are divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths; according to the screened task paths, an optimal decision plan is obtained, and the optimal decision plan is executed to dispatch the large power grid, thereby improving the efficiency and accuracy of judging the safety of the electric power system and the overall process of electric power dispatching.

Description

Technical Field

[0001] The present application relates to the technical field of electric power dispatching, and in particular to a large power grid dispatching method and intelligent agent model adapted to an electric power digital simulation system. Background Art

[0002] Traditional power system dispatch can be defined as modeling each basic unit in the target power grid from the system level, determining the optimization variables, constraints and goals in the dispatch optimization model, and then finding the optimal solution that meets the constraints under the guidance of the goal. This is a dispatch optimization method guided by this type of theoretical framework. However, with the rapid development of the power system, the physical characteristics of the power generation side and the power consumption side have gradually diversified. The large-scale application of new energy such as wind and solar power and the massive number of dispatchable units have made the dispatch optimization problem more and more complicated. The limitations of traditional model-based dispatch optimization methods have gradually become apparent.

[0003] At present, the research on intelligent dispatching methods for auxiliary decision-making is mostly theoretical or can only achieve semi-automatic decision-making, and it is impossible to completely get rid of human intervention in the process of power dispatch optimization. Power dispatching begins with manual dispatching, that is, the dispatcher manually issues dispatching instructions after a series of calculations and verifications. However, manual dispatching often has many problems such as huge calculation workload, slow dispatch response, low dispatch timeliness and high dispatch error rate. Summary of the invention

[0004] The present application provides a large power grid dispatching method and intelligent agent model adapted to a power digital simulation system, so as to improve the efficiency and accuracy of judging the safety of the power system and the overall process of power dispatching.

[0005] The present application provides a large power grid dispatching method adapted to a power digital simulation system, comprising:

[0006] The grid topology data of the power system is input into the intelligent prediction model, and decisions are made on the preset problems to obtain the decision tasks, which are as follows:

[0007] Inputting the power grid topology data into an intelligent prediction model, wherein the intelligent prediction model is a graph convolutional neural network decision model, so that the intelligent prediction model inputs the power grid topology data and the expert data set into the graph convolutional network and obtains a decision task through a linear layer;

[0008] The preset problem includes: one of a safety assessment problem and a fault location problem; the intelligent prediction model is obtained by training historical simulation data; the historical simulation data is obtained by simulating a power digital simulation system;

[0009] The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, including:

[0010] If the preset problem is not fault location, the communication consumption and execution consumption of the decision task are used as sample features to divide the decision task into several task branches;

[0011] Using a plurality of the task branches as decision branches of a decision tree to establish a decision tree;

[0012] Performing a safety assessment on each of the decision branches, determining a decision branch with a high risk of failure, and performing a pruning operation on the decision branch with a high risk of failure to obtain a pruned decision tree;

[0013] The optimal decision-making solution is obtained according to the task path after screening, including:

[0014] According to the pruned decision tree, the analysis data is obtained;

[0015] The analysis data includes: state estimation report, power flow analysis report, reactive power optimization report and safety analysis report; the power flow analysis report includes: unreasonable results of system voltage and frequency in the current state; the reactive power optimization report includes: improving voltage quality to reduce network losses; the static safety analysis report includes: what state the system is in, which components’ overload will affect the safety of the system, and the safety recovery strategy of the system in the current state;

[0016] Entering the analysis data into a database to update the tidal current data;

[0017] Calculate the optimal decision plan using updated power flow data;

[0018] The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, further comprising:

[0019] If the preset problem is fault location, the decision-making task is divided to generate several task paths;

[0020] Performing fault analysis on several of the task paths, the fault analysis includes: determining whether the switch and automatic safety device are operating correctly, determining whether the fault is caused by human error or malfunction, including: the location, type and nature of the fault, and preliminary fault handling suggestions, including: determining whether the component status and related equipment are available;

[0021] Among all task paths, select a task path that minimizes system loss as the selected task path;

[0022] Obtaining an optimal decision plan according to the screened task paths, and executing the optimal decision plan to dispatch the large power grid;

[0023] The optimal decision-making plan includes: processing plans for components that need to be isolated, power outage areas that need to be restored, and equipment that needs to be suspended, and the record and explanation of the status of each running software during the fault recovery process of the equipment suspension processing plan;

[0024] The step of obtaining the optimal decision solution according to the task path after screening also includes:

[0025] The task path that minimizes power system losses is taken as the optimal decision-making solution.

[0026] Furthermore, after obtaining the optimal decision solution according to the task path after screening, the method further includes:

[0027] Retrieving task execution data, the task execution data including: task path, execution time, resource consumption and fault monitoring results;

[0028] Retrieving information from an external data source, wherein the information from the external data source includes: historical monitoring data, an expert database, and real-time monitoring data;

[0029] The task execution data and the information of the external data source are submitted to the user.

[0030] Further, monitoring resources in the power system and the operating status of said resources;

[0031] If a computing resource is occupied but not running within a preset time, or a computing resource is not supported, the running status will be fed back.

[0032] Furthermore, in response to the user's data request operation, the corresponding data is sent to the external network through the physical isolation layer for the user to use;

[0033] In the process of inputting the grid topology data of the power system into the intelligent prediction model and making decisions on preset problems, if it is determined that the power system meets the preset conditions, the operation data of the power system is notified to the user through the human-machine interface;

[0034] The preset conditions include: the operation data of the power system reaches a fault threshold.

[0035] As a preferred solution, this application conducts offline training on the data sets of power grid topology data and historical expert decision data, and deploys the decision model online after obtaining it. It responds to power dispatching needs and sends the task to the decision module. The decision task derived by the decision module makes full use of the structural information of the network topology. At the same time, combined with the expert data set, it can accurately obtain the decision mapping from network node state to state, thereby improving the efficiency and accuracy of decision generation.

[0036] Accordingly, the present application also provides a large power grid dispatching intelligent agent model device adapted to the power digital simulation system, including: a decision module and a management module;

[0037] The decision module is used to input the power grid topology data into the intelligent prediction model, make decisions on preset problems, and obtain decision tasks, which are specifically:

[0038] Inputting the power grid topology data into an intelligent prediction model, wherein the intelligent prediction model is a graph convolutional neural network decision model, so that the intelligent prediction model inputs the power grid topology data and the expert data set into the graph convolutional network and obtains a decision task through a linear layer;

[0039] The preset problem includes: one of a safety assessment problem and a fault location problem; the intelligent prediction model is obtained by training historical simulation data; the historical simulation data is obtained by simulating a power digital simulation system;

[0040] The management module is used to divide the decision-making task to obtain a plurality of task paths, and analyze and screen the plurality of task paths to obtain screened task paths, including:

[0041] If the preset problem is not fault location, the communication consumption and execution consumption of the decision task are used as sample features to divide the decision task into several task branches;

[0042] Using a plurality of the task branches as decision branches of a decision tree to establish a decision tree;

[0043] Performing a safety assessment on each of the decision branches, determining a decision branch with a high risk of failure, and performing a pruning operation on the decision branch with a high risk of failure to obtain a pruned decision tree;

[0044] The optimal decision-making solution is obtained according to the task path after screening, including:

[0045] According to the pruned decision tree, the analysis data is obtained;

[0046] The analysis data includes: state estimation report, power flow analysis report, reactive power optimization report and safety analysis report; the power flow analysis report includes: unreasonable results of system voltage and frequency in the current state; the reactive power optimization report includes: improving voltage quality to reduce network losses; the static safety analysis report includes: what state the system is in, which components’ overload will affect the safety of the system, and the safety recovery strategy of the system in the current state;

[0047] Entering the analysis data into a database to update the tidal current data;

[0048] Calculate the optimal decision plan using updated power flow data;

[0049] The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, further comprising:

[0050] If the preset problem is fault location, the decision-making task is divided to generate several task paths;

[0051] Performing fault analysis on several of the task paths, the fault analysis includes: determining whether the switch and automatic safety device are operating correctly, determining whether the fault is caused by human error or malfunction, including: the location, type and nature of the fault, and preliminary fault handling suggestions, including: determining whether the component status and related equipment are available;

[0052] Among all task paths, select a task path that minimizes system loss as the selected task path;

[0053] Obtaining an optimal decision plan according to the screened task paths, and executing the optimal decision plan to dispatch the large power grid;

[0054] The optimal decision-making plan includes: processing plans for components that need to be isolated, power outage areas that need to be restored, and equipment that needs to be suspended, and the record and explanation of the status of each running software during the fault recovery process of the equipment suspension processing plan;

[0055] The step of obtaining the optimal decision solution according to the task path after screening also includes:

[0056] The task path that minimizes power system losses is taken as the optimal decision-making solution.

[0057] Accordingly, the present application also provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute a large power grid dispatching method adapted to the power digital simulation system as described in the content of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flow chart of an embodiment of a large power grid dispatching method adapted to a power digital simulation system provided by the present application;

[0059] Figure 2 It is a structural schematic diagram of an embodiment of a large power grid dispatching intelligent body model device adapted to the power digital simulation system provided by the present application. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0061] Embodiment 1

[0062] Please refer to Figure 1 , a large power grid dispatching method adapted to a power digital simulation system provided in an embodiment of the present application, applied to a large power grid dispatching intelligent agent model device adapted to a power digital simulation system, the model comprising: a decision module, a management module, an information module, a monitoring module and a user interface module;

[0063] The decision module generates the main agent for scheduling task plans (decision plans).

[0064] The management module is the core of the entire system model. It stores other information in the system. It is mainly responsible for decision-making, querying, controlling and supervising other issues. It coordinates members of different project teams. The management module formulates task planning, scheduling, decomposition, etc. according to the requirements of users and consults with the management of other subsystems.

[0065] The intelligent decision-making method comprises steps S101-S103:

[0066] Step S101: inputting the grid topology data of the power system into an intelligent prediction model, making decisions on preset problems, and obtaining decision tasks; the intelligent prediction model is trained by historical simulation data; the historical simulation data is simulated by a power digital simulation system;

[0067] Furthermore, the intelligent prediction model is a graph convolutional neural network decision model.

[0068] Furthermore, the grid topology data of the power system is input into the intelligent prediction model, and a decision is made on a preset problem to obtain a decision task, which is specifically:

[0069] Inputting the power grid topology data into the intelligent prediction model, so that the intelligent prediction model inputs the power grid topology data and the expert data set into the graph convolutional network, and through the linear layer, obtains the decision task;

[0070] The preset question includes: one of a safety assessment question and a fault location question.

[0071] In this embodiment, the graph convolutional neural network (GNN) in the graph neural network (GNN) is used to construct the intelligent prediction model. The graph convolutional neural network model framework GCN is a first-order approximation of ChebyNet, that is, the first-order neighbor features of the central node are transformed and aggregated to its central node. While taking into account the graph topology data, the algorithm optimizes the previous problems of high computational complexity and data distortion. Its feature iteration formula is:

[0072] ;

[0073] ;

[0074] in, is the feature matrix, is the adjacency matrix of the graph nodes, is the degree matrix of the node.

[0075] The intelligent prediction model adopts the mode of offline learning plus online monitoring. The construction of the training feature set will be constructed using historical simulation data. The data set is obtained for offline training, and the intelligent prediction model is deployed online. The management module responds to the dispatching needs of the power system and sends the task to the decision module, which then comes up with a decision plan.

[0076] Suppose the power grid graph data at a certain moment is G{X, E}, where X is the power flow information matrix of several nodes, and E is the record matrix of the connection between power grid nodes. Suppose the expert dispatch data at this historical moment is Y, which also contains the decision information of each node. The power grid graph data G{X, E} is the power grid topology data, and the expert dispatch data is the expert data set. The power grid graph data and the expert dispatch data are input into the graph convolutional network, and through the linear layer, several main decision agents of the dispatch task plan are generated, that is, decision tasks.

[0077] In one preferred implementation, power software such as power flow analysis is connected to the decision module, which has a steady-state analysis function. The steady-state analysis is used to evaluate the system status, and there are two startup modes: manual and automatic. When a power network failure occurs in the jurisdiction and a steady-state analysis is required, the dispatcher can manually run the steady-state analysis module as needed, or define a fixed time or certain events to trigger the steady-state analysis.

[0078] Step S102: dividing the decision task to obtain a plurality of task paths, and analyzing and screening the plurality of task paths to obtain screened task paths;

[0079] Step S103: Obtain an optimal decision plan based on the screened task paths, and execute the optimal decision plan to dispatch the large power grid.

[0080] Furthermore, the decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, including:

[0081] If the preset problem is not fault location, the communication consumption and execution consumption of the decision task are used as sample features to divide the decision task into several task branches;

[0082] Using a plurality of the task branches as decision branches of a decision tree to establish a decision tree;

[0083] A safety assessment is performed on each of the decision branches to determine a decision branch with a high risk of failure, and a pruning operation is performed on the decision branch with a high risk of failure to obtain a pruned decision tree.

[0084] Furthermore, the optimal decision-making solution is obtained according to the task path after screening, specifically:

[0085] According to the pruned decision tree, the analysis data is obtained;

[0086] Entering the analysis data into a database to update the tidal current data;

[0087] Utilize the updated power flow data to calculate the optimal decision-making plan.

[0088] In this embodiment, since the management module is responsible for the overall planning of the entire system, a machine learning algorithm is also required to make it intelligent. This embodiment uses a decision tree model with fault penalties to model the task management function of the management module.

[0089] Decision Tree is a model that uses a tree data structure to display decision rules and classification results. As an inductive learning algorithm, its focus is to transform seemingly disordered and messy known data into a tree model that can predict unknown data through some technical means. Each path from the root node (the attribute that contributes the most to the final classification result) to the leaf node (the final classification result) represents a decision rule. The basis for dividing leaf nodes is usually divided by information gain or Gini coefficient. Usually, if you only construct a decision tree and allow it to divide nodes, it will result in a high accuracy rate on the training set, but poor effect on the test set, overfitting of the model, and weak generalization ability. Therefore, we need pruning to prevent the decision tree from overfitting and remove the node branches that should not be divided.

[0090] This embodiment uses the communication consumption and execution consumption of the management module task as sample features, and uses the traversal method to divide the decision tasks generated by the decision module, thereby establishing a decision tree. At the same time, a safety assessment is performed on the formed task branches, and the expert data set is used to prune the task operation decision branches with high failure risks. Assume that there is a task set {A, B, ..., I}, and after the decision tree is divided, a variety of task paths {ABD, ABEH, ABEI, ABF, ACG} are formed. At this time, for each possible path, it will be checked through the fault monitoring function containing the expert data set, including: judging whether it complies with the operating procedures and operations of the power system, and system stability evaluation. Assume that the task path of ABEI does not meet the safety assessment after verification by the expert data set, so the branch is eliminated, and the remaining execution paths after pruning are executed according to the current decision division.

[0091] In one preferred embodiment, in terms of communication consumption characteristics, steady-state analysis involves a large amount of data exchange, such as transmission of network topology, real-time status data, etc. The time required to transmit these data can be used as a characteristic of communication consumption; in terms of execution consumption characteristics, the calculation time required for each step in steady-state analysis (such as power flow calculation, reactive power optimization, and static safety analysis, etc.) can be used as a characteristic of execution consumption.

[0092] The decision-making tasks of steady-state analysis involve advanced applications such as network topology, state estimation, power flow calculation, reactive power optimization, static safety analysis and load forecasting. Through network topology analysis, the operation of each device in the system is understood, and the physical model in the network topology is converted into a calculation model. A safety assessment is performed on each of the decision branches, and the transmitted data is corrected and erroneous data is deleted through state estimation. After the state estimation, power flow calculation, reactive power optimization and static safety analysis are completed, analysis data such as state estimation report, power flow analysis report, reactive power optimization report and safety analysis report will be generated respectively.

[0093] Entering the analysis data into a database to update the tidal current data;

[0094] Utilize the updated power flow data to calculate the optimal decision-making plan.

[0095] Among them, the operating status of the system is calculated through power flow calculation.

[0096] The power flow analysis report gives unreasonable results such as system voltage and frequency in the current state. The reactive power optimization report takes into account the economy of the system while ensuring the normal operation of the system. For example, improving the quality of voltage to reduce network losses. The static safety analysis report includes the state of the system, which components' overload will affect the safety of the system, and gives the safety recovery strategy for the state. The dispatcher only needs to observe and receive whether the operation of the intelligent dispatching system is normal and decide whether to trust the decision of the intelligent dispatching system.

[0097] Furthermore, the dividing the decision task to obtain a plurality of task paths, and analyzing and screening the plurality of task paths to obtain screened task paths, further includes:

[0098] If the preset problem is fault location, the decision-making task is divided to generate several task paths;

[0099] Among all task paths, select a task path that minimizes system loss as the screened task path.

[0100] Furthermore, the optimal decision-making solution is obtained according to the task path after screening, specifically:

[0101] The task path that minimizes power system losses is taken as the optimal decision-making solution.

[0102] In one preferred implementation, different from the safety assessment of the management module when forming the task path decision tree, when a real fault occurs, the decision module is called to perform fault analysis and decision-making; if it is a fault, a decision plan is formed; the task path is generated through the management module, and the task path screening is performed, and the task path screening should be changed to minimize system losses. Faults include fault diagnosis, fault recovery and fault operation tickets.

[0103] If the decision module fault analysis and decision determines that it is not a fault, the abnormal data input into the decision module will be traced.

[0104] The decision-making module performs fault analysis including: judging whether the switch and automatic safety device are operating correctly or not, judging whether the fault is caused by human error or malfunction, including the location, type and nature of the fault, giving preliminary fault handling opinions, including component status and the availability of related equipment, etc.

[0105] The decision-making module makes fault decisions including: providing an overall safety recovery plan, including solutions for components that need to be isolated, power outage areas that need to be restored, and equipment that needs to be suspended, as well as the record and explanation of the status of each running software during the fault recovery process of the equipment suspension processing plan.

[0106] In addition, for the sake of system security, the command and coordination layer has set the priority of fault analysis to the highest level. When a fault is diagnosed, the fault handling should be coordinated with other tasks to maximize the use of system performance resources for fault analysis applications.

[0107] Furthermore, after obtaining the optimal decision solution according to the task path after screening, the method further includes:

[0108] Retrieving task execution data, the task execution data including: task path, execution time, resource consumption and fault monitoring results;

[0109] Retrieving information from an external data source, wherein the information from the external data source includes: historical monitoring data, an expert database, and real-time monitoring data;

[0110] The task execution data and the information of the external data source are submitted to the user.

[0111] In this embodiment, the information module is used to retrieve the task execution data and the information of the external data source, and submit them to the user; after submitting the task execution data and the information of the external data source to the user, it also includes: feedback and sorting of some problems based on the retrieved data.

[0112] Further, monitoring resources in the power system and the operating status of said resources;

[0113] If a computing resource is occupied but not running within a preset time, or a computing resource is not supported, the running status will be fed back.

[0114] In this embodiment, the monitoring module is used to monitor the resources in the system and the operation status of each Agent to prevent the computing resources from being occupied for a long time but not running, or the situation where a large amount of computing resources are needed but not enough support is obtained. The system status can be fed back to the management module to ensure the normal and stable operation of the system.

[0115] Furthermore, in response to the user's data request operation, the corresponding data is sent to the external network through the physical isolation layer for the user to use;

[0116] In the process of inputting the grid topology data of the power system into the intelligent prediction model and making decisions on preset problems, if it is determined that the power system meets the preset conditions, the operation data of the power system is notified to the user through the human-machine interface;

[0117] The preset conditions include: the operation data of the power system reaches a fault threshold.

[0118] In this embodiment, the user interface module is used to provide friendly interaction for two-way communication between people and machines through the interface. The main task of the agent is to obtain information from the user. The agent provides the user with conclusions and explanation mechanisms. The agent asks the user to input necessary information when solving problems. This part increases the intelligence of the publishing system. In addition to sending the data of the intranet to the external network through the physical isolation layer for the dispatcher to use, it can also actively inform the user of the system's operating status through the browser under given conditions. For example, when a major fault occurs, the browser of the client responsible for the fault will be forced to open to remind the user of the occurrence of the fault. This greatly improves the speed of the dispatch department's response to faults and its timely coordination ability.

[0119] Implementing the embodiments of the present application has the following effects:

[0120] This application conducts offline training on the data set of power grid topology data and historical expert decision data, obtains the decision model and deploys it online, responds to power dispatching needs, sends tasks to the decision module, and the decision module obtains the decision task, making full use of the structural information of the network topology. At the same time, combined with the expert data set, it can accurately obtain the decision mapping from the state of the network node to the state, thereby improving the efficiency and accuracy of decision generation;

[0121] This application proposes an intelligent decision-making method in which multiple agents work together. Combined with the feedback from users, monitoring, and information agents on the optimization of decision-making plans, the management module optimizes the decision-making plans, judges the safety of the power system, and dispatches the overall process, further improving the efficiency and accuracy of judging the safety of the power system and the overall process of power dispatching.

[0122] Embodiment 2

[0123] Please refer to Figure 2 , a large power grid dispatching intelligent agent model device adapted to the power digital simulation system provided in the embodiment of the present application, including: a decision module 201, a management module 202, an information module 203, a monitoring module 204 and a user interface module 205;

[0124] The decision module 201 is used to input the power grid topology data into the intelligent prediction model, make decisions on preset problems, and obtain decision tasks, which are specifically:

[0125] Inputting the power grid topology data into an intelligent prediction model, wherein the intelligent prediction model is a graph convolutional neural network decision model, so that the intelligent prediction model inputs the power grid topology data and the expert data set into the graph convolutional network and obtains a decision task through a linear layer;

[0126] The preset problem includes: one of a safety assessment problem and a fault location problem; the intelligent prediction model is obtained by training historical simulation data; the historical simulation data is obtained by simulating a power digital simulation system;

[0127] In this embodiment, the decision-making module will decide whether to accept the task alone or complete the task with others according to its own capabilities. It will include a decision resource layer, which is composed of the power network database, agent model library and expert database of the entire system, and provides data agent model and knowledge support for upper-level applications. This part includes a lot of advanced application software, such as flow calculation, reactive power optimization and static safety analysis, etc. It mainly uses the real-time data of the data platform to analyze the power system, obtain the current operating status of the power system, and provide decision support for the dispatcher. Since the intelligent decision-making system has high requirements on the timeliness of data, the first-level data storage uses a real-time database, and the second-level storage uses a database management system. The data of the data platform comes from the remote signal and remote control data of each plant and station, and the database can also be updated as required.

[0128] The management module 202 is used to divide the decision-making task to obtain a plurality of task paths, and analyze and filter the plurality of task paths to obtain filtered task paths, including:

[0129] If the preset problem is not fault location, the communication consumption and execution consumption of the decision task are used as sample features to divide the decision task into several task branches;

[0130] Using a plurality of the task branches as decision branches of a decision tree to establish a decision tree;

[0131] Performing a safety assessment on each of the decision branches, determining a decision branch with a high risk of failure, and performing a pruning operation on the decision branch with a high risk of failure to obtain a pruned decision tree;

[0132] The optimal decision-making solution is obtained according to the task path after screening, including:

[0133] According to the pruned decision tree, the analysis data is obtained;

[0134] The analysis data includes: state estimation report, power flow analysis report, reactive power optimization report and safety analysis report; the power flow analysis report includes: unreasonable results of system voltage and frequency in the current state; the reactive power optimization report includes: improving voltage quality to reduce network losses; the static safety analysis report includes: what state the system is in, which components’ overload will affect the safety of the system, and the safety recovery strategy of the system in the current state;

[0135] Entering the analysis data into a database to update the tidal current data;

[0136] Calculate the optimal decision plan using updated power flow data;

[0137] The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, further comprising:

[0138] If the preset problem is fault location, the decision-making task is divided to generate several task paths;

[0139] Performing fault analysis on several of the task paths, the fault analysis includes: determining whether the switch and automatic safety device are operating correctly, determining whether the fault is caused by human error or malfunction, including: the location, type and nature of the fault, and preliminary fault handling suggestions, including: determining whether the component status and related equipment are available;

[0140] Among all task paths, select a task path that minimizes system loss as the selected task path;

[0141] Obtaining an optimal decision plan according to the screened task paths, and executing the optimal decision plan to dispatch the large power grid;

[0142] The optimal decision-making plan includes: processing plans for components that need to be isolated, power outage areas that need to be restored, and equipment that needs to be suspended, and the record and explanation of the status of each running software during the fault recovery process of the equipment suspension processing plan;

[0143] The step of obtaining the optimal decision solution according to the task path after screening also includes:

[0144] The task path that minimizes power system losses is taken as the optimal decision-making solution.

[0145] In this embodiment, the management module is the core part of the entire system model. Agents store other information in the system. Agents are mainly responsible for decision-making, querying, and controlling and supervising other agents. Agents coordinate members of different project teams. The main problem in the process is task decomposition and allocation and negotiation between them. Management formulates task planning and scheduling, decomposition, etc. according to the requirements of users and negotiates with management in other subsystems.

[0146] The information module 203 is used to retrieve task execution data, and the task execution data includes: task path, execution time, resource consumption and fault monitoring results;

[0147] Retrieving information from an external data source, wherein the information from the external data source includes: historical monitoring data, an expert database, and real-time monitoring data;

[0148] Submitting the task execution data and the information of the external data source to the user;

[0149] In this embodiment, the information module is responsible for processing the retrieved information and then submitting the information to the user, and then providing feedback and sorting out some issues on this basis.

[0150] The monitoring module 204 is used to monitor the resources in the power system and the operating status of the resources;

[0151] If a situation occurs where the computing resources are occupied but not running within the preset time, or if a situation occurs where the computing resources cannot be supported, the running status will be fed back;

[0152] In this embodiment, the monitoring module is mainly responsible for monitoring the resources in the system and the operation status of each Agent to prevent the computing resources from being occupied for a long time but not running, or the situation where a large amount of computing resources are needed but not enough support is obtained. It can actually feedback the system status to the management module to ensure the normal and stable operation of the system.

[0153] The user interface module 205 is used to respond to the user's data request operation and send the corresponding data to the external network through the physical isolation layer for the user to use;

[0154] In the process of inputting the grid topology data of the power system into the intelligent prediction model and making decisions on preset problems, if it is determined that the power system meets the preset conditions, the operation data of the power system is notified to the user through the human-machine interface;

[0155] The preset conditions include: the operation data of the power system reaches a fault threshold.

[0156] In this embodiment, the user interface module provides friendly interaction between humans and machines through an interface. The main task of the agent is to obtain information from the user. The agent provides conclusions and explanation mechanisms to the user. The agent requests the user to input necessary information when solving problems.

[0157] The above-mentioned large power grid dispatching intelligent agent model adapted to the power digital simulation system can implement the large power grid dispatching method adapted to the power digital simulation system of the above-mentioned method embodiment. The options in the above-mentioned method embodiment are also applicable to this embodiment and will not be described in detail here. The rest of the contents of the embodiments of this application can refer to the contents of the above-mentioned method embodiment, and will not be repeated in this embodiment.

[0158] Embodiment 3

[0159] Accordingly, the present application also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the large power grid dispatching method of the adaptive power digital simulation system as described in any one of the above embodiments.

[0160] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the terminal device.

[0161] The terminal device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0162] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and uses various interfaces and lines to connect various parts of the entire terminal device.

[0163] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the terminal device by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function, etc.; the data storage area can store data created according to the use of the mobile terminal, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0164] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0165] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A large power grid dispatching method adapted to a power digital simulation system, characterized in that: include: The grid topology data of the power system is input into the intelligent prediction model, and decisions are made on the preset problems to obtain the decision tasks, which are as follows: Inputting the power grid topology data into an intelligent prediction model, wherein the intelligent prediction model is a graph convolutional neural network decision model, so that the intelligent prediction model inputs the power grid topology data and the expert data set into the graph convolutional network and obtains a decision task through a linear layer; The preset problem includes: one of a safety assessment problem and a fault location problem; the intelligent prediction model is obtained by training historical simulation data; the historical simulation data is obtained by simulating a power digital simulation system; The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, including: If the preset problem is not fault location, the communication consumption and execution consumption of the decision task are used as sample features to divide the decision task into several task branches; Using a plurality of the task branches as decision branches of a decision tree to establish a decision tree; Performing a safety assessment on each of the decision branches, determining a decision branch with a high risk of failure, and performing a pruning operation on the decision branch with a high risk of failure to obtain a pruned decision tree; According to the task path after screening, the optimal decision-making plan is obtained, including: According to the pruned decision tree, the analysis data is obtained; The analysis data includes: state estimation report, power flow analysis report, reactive power optimization report and safety analysis report; the power flow analysis report includes: unreasonable results of system voltage and frequency in the current state; the reactive power optimization report includes: improving voltage quality to reduce network losses; the static safety analysis report includes: what state the system is in, which components’ overload will affect the safety of the system, and the safety recovery strategy of the system in the current state; Entering the analysis data into a database to update the tidal current data; Utilizing the updated power flow data, calculating the optimal decision-making scheme, and executing the optimal decision-making scheme to dispatch the large power grid; The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, further comprising: If the preset problem is fault location, the decision-making task is divided to generate several task paths; Performing fault analysis on several of the task paths, the fault analysis includes: determining whether the switch and automatic safety device are operating correctly, determining whether the fault is caused by human error or malfunction, including: the location, type and nature of the fault, and preliminary fault handling suggestions, including: determining whether the component status and related equipment are available; Among all task paths, select a task path that minimizes system loss as the selected task path; Obtaining an optimal decision plan according to the screened task paths, and executing the optimal decision plan to dispatch the large power grid; The optimal decision plan includes: processing plans for components that need to be isolated, power outage areas that need to be restored, and equipment that needs to be suspended, and the record and explanation of the status of each running software during the fault recovery process of the equipment suspension processing plan.

2. A large power grid dispatching method adapted to a power digital simulation system as claimed in claim 1, characterized in that: After obtaining the optimal decision solution according to the task path after screening, the method further includes: Retrieving task execution data, the task execution data including: task path, execution time, resource consumption and fault monitoring results; Retrieving information from an external data source, wherein the information from the external data source includes: historical monitoring data, an expert database, and real-time monitoring data; The task execution data and the information of the external data source are submitted to the user.

3. A large power grid dispatching method adapted to a power digital simulation system as claimed in claim 1, characterized in that: Also includes: Monitoring resources in the power system and the operating status of said resources; If a computing resource is occupied but not running within a preset time, or a computing resource is not supported, the running status will be fed back.

4. A large power grid dispatching method adapted to a power digital simulation system as claimed in claim 1, characterized in that: Also includes: Respond to the user's data request operation and send the corresponding data through the physical isolation layer to the external network for the user to use; In the process of inputting the grid topology data of the power system into the intelligent prediction model and making decisions on preset problems, if it is determined that the power system meets the preset conditions, the operation data of the power system is notified to the user through the human-machine interface; The preset conditions include: the operation data of the power system reaches a fault threshold.

5. A large power grid dispatching intelligent agent model device adapted to the power digital simulation system, characterized in that: include: Decision-making module and management module; The decision module is used to input the power grid topology data into the intelligent prediction model, make decisions on preset problems, and obtain decision tasks, which are specifically: Inputting the power grid topology data into an intelligent prediction model, wherein the intelligent prediction model is a graph convolutional neural network decision model, so that the intelligent prediction model inputs the power grid topology data and the expert data set into the graph convolutional network and obtains a decision task through a linear layer; The preset problem includes: one of a safety assessment problem and a fault location problem; the intelligent prediction model is obtained by training historical simulation data; the historical simulation data is obtained by simulating a power digital simulation system; The management module is used to divide the decision-making task to obtain a plurality of task paths, and analyze and screen the plurality of task paths to obtain screened task paths, including: If the preset problem is not fault location, the communication consumption and execution consumption of the decision task are used as sample features to divide the decision task into several task branches; Using a plurality of the task branches as decision branches of a decision tree to establish a decision tree; Performing a safety assessment on each of the decision branches, determining a decision branch with a high risk of failure, and performing a pruning operation on the decision branch with a high risk of failure to obtain a pruned decision tree; According to the task path after screening, the optimal decision-making plan is obtained, including: According to the pruned decision tree, the analysis data is obtained; The analysis data includes: state estimation report, power flow analysis report, reactive power optimization report and safety analysis report; the power flow analysis report includes: unreasonable results of system voltage and frequency in the current state; the reactive power optimization report includes: improving voltage quality to reduce network losses; the static safety analysis report includes: what state the system is in, which components’ overload will affect the safety of the system, and the safety recovery strategy of the system in the current state; Entering the analysis data into a database to update the tidal current data; Utilizing the updated power flow data, calculating the optimal decision-making scheme, and executing the optimal decision-making scheme to dispatch the large power grid; The decision-making task is divided to obtain a plurality of task paths, and the plurality of task paths are analyzed and screened to obtain screened task paths, further comprising: If the preset problem is fault location, the decision-making task is divided to generate several task paths; Performing fault analysis on several of the task paths, the fault analysis includes: determining whether the switch and automatic safety device are operating correctly, determining whether the fault is caused by human error or malfunction, including: the location, type and nature of the fault, and preliminary fault handling suggestions, including: determining whether the component status and related equipment are available; Among all task paths, select a task path that minimizes system loss as the selected task path; Obtaining an optimal decision plan according to the screened task paths, and executing the optimal decision plan to dispatch the large power grid; The optimal decision plan includes: processing plans for components that need to be isolated, power outage areas that need to be restored, and equipment that needs to be suspended, and the record and explanation of the status of each running software during the fault recovery process of the equipment suspension processing plan.

Images