High-efficiency autonomous controllable electromagnetic transient real-time simulation method and simulation platform

Abstract

A high-efficiency autonomous controllable electromagnetic transient real-time simulation method and simulation platform relate to the field of electrical digital data processing, and the method comprises the following steps: constructing a space-time diagram neural network model according to power grid equipment and a connection relationship thereof, inputting physical parameters of the power grid equipment into node feature vectors, calculating a fault signal propagation time matrix, and calculating a fault signal propagation time matrix; power grid nodes are dynamically layered and divided into a core layer, a near-domain layer, a transition layer and a far-domain layer, the core layer is close to a fault point, basic step length calculation is adopted for the core layer, for example, large time step length calculation is adopted for the far-domain layer far away from the fault point, and electrical characteristic parameters of the fault nodes are calculated in the core layer to recognize a key node set. And taking interlayer boundary node parameters in the key node set as adjacent layer calculation boundary conditions, and outputting transient process data of each power grid node, dynamic change data of a fault influence range and parameter alarm information of the key node. By implementing the method, the computing resource consumption of electromagnetic transient real-time simulation is reduced.

Description

Technical Field

[0001] This application relates to the field of electrical digital data processing, and in particular, to a high-performance, self-controlled electromagnetic transient real-time simulation method and simulation platform. Background Art

[0002] With the continuous expansion of the scale of the power system and the continuous increase in the proportion of new energy grid connection, the complexity and uncertainty of power grid operation have increased significantly, and the importance of electromagnetic transient characteristic analysis has become increasingly prominent. Electromagnetic transient real-time simulation, as an important means to study the dynamic characteristics of the power grid and evaluate the system stability, is of great significance for ensuring the safe and stable operation of the power grid.

[0003] In the related art, the mathematical model of power grid equipment can be established, and the numerical integration algorithm can be used to solve the system state. In the implementation of this method, first, the equivalent models of each component of the power grid are constructed, then the network equations are established by using the node voltage method, and finally, the numerical calculation method is used to perform iterative operations on the entire power grid system to obtain the transient response process of each node.

[0004] However, due to the large scale of the power grid and the large number of nodes, when performing the transient calculation of the entire network, continuous numerical iterative operations need to be performed on all nodes. Especially in a large-scale power grid system, the computational amount increases exponentially with the number of nodes, resulting in a large consumption of computing resources. Summary of the Invention

[0005] This application provides a high-performance, self-controlled electromagnetic transient real-time simulation method and simulation platform for reducing the consumption of computing resources for electromagnetic transient real-time simulation.

[0006] In a first aspect, the present application provides a high-performance, autonomous and controllable electromagnetic transient real-time simulation method, which is applied to an electromagnetic transient real-time simulation platform. The method includes: constructing a spatio-temporal graph neural network model according to power grid devices and the physical connection relationships between the power grid devices. The spatio-temporal graph neural network model constructs the power grid devices as power grid nodes and constructs the physical connection relationships between the power grid devices as network connection relationships; inputting the physical parameters of the power grid devices into the feature vectors of the network nodes to obtain a propagation time matrix of the fault signal from the fault point to each power grid node; dynamically stratifying each power grid node according to the propagation time matrix, dividing the power grid nodes into a core layer, a near-domain layer, a transition layer, and a far-domain layer according to the propagation time, and calculating the transient processes of the power grid nodes in each layer with different time steps respectively to obtain the transient process data of each power grid node; calculating the electrical characteristic parameters of the fault node in the core layer, identifying a set of key nodes, and using the parameters of the inter-layer boundary nodes in the set of key nodes as the boundary conditions for adjacent layer calculations; inputting the transient process data into a deep reinforcement learning model to obtain a calculation frequency adjustment coefficient for each power grid node; if the calculation frequency adjustment coefficient of a target node is greater than a preset threshold, upgrading the target node to an upper-layer node for transient calculation, and the calculation time step of the upper-layer node is smaller than that of the target node; in response to a user's data export operation, outputting the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes.

[0007] By adopting the above technical solution, a spatio-temporal graph neural network model is constructed to present the power grid structure. The propagation time matrix is obtained by inputting physical parameters. Layers are divided accordingly. The core layer is calculated with a basic time step, and the far-domain layer is calculated with a larger time step, reasonably allocating computing resources, reducing consumption, calculating electrical characteristic parameters in the core layer to identify a set of key nodes, and using the parameters of their boundary nodes as boundary conditions to ensure accurate calculation. Finally, accurate transient process data is obtained, improving the calculation efficiency.

[0008] Combined with some embodiments of the first aspect, in some embodiments, the step of dynamically stratifying each power grid node according to the propagation time matrix, dividing the power grid nodes into a core layer, a near-domain layer, a transition layer, and a far-domain layer according to the propagation time, and calculating with different time steps respectively specifically includes: stratifying the power grid nodes according to the propagation time matrix, dividing the nodes with a propagation time less than T into the core layer and calculating once every basic time step, dividing the nodes with a propagation time between T and 2T into the near-domain layer and calculating once every two times the basic time step, dividing the nodes with a propagation time between 2T and 4T into the transition layer and calculating once every four times the basic time step, dividing the nodes with a propagation time greater than 4T into the far-domain layer and calculating once every eight times the basic time step to obtain the transient process data of each power grid node.

[0009] By adopting the above technical solution, stratifying according to the propagation time matrix, the core layer calculates at a high frequency with a basic step size to ensure accuracy, and the far-field layer calculates at a low frequency with eight times the basic step size to reduce resource consumption. Intermediate layers such as the near-field layer and the transition layer also adopt corresponding step sizes according to the propagation time, reducing the calculation amount while ensuring a certain accuracy. Overall, a reasonable allocation of computing resources is achieved, improving the computing efficiency while ensuring the accuracy of key areas, and efficiently obtaining transient process data.

[0010] Combined with some embodiments of the first aspect, in some embodiments, the step of outputting the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter alarm information of the key nodes in response to the user's data export operation specifically includes: in response to the user's data export operation, obtaining the voltage drop, overcurrent, and power fluctuation parameters of each layer of power grid nodes, generating parameter alarm information according to a preset threshold; generating dynamic change data of the fault influence range based on the propagation time matrix; and outputting the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter alarm information of the key nodes.

[0011] By adopting the above technical solution, when responding to the export operation, obtaining parameters to generate alarm information can promptly detect abnormalities. Generating dynamic change data of the fault influence range based on the propagation time matrix to display the fault propagation process. Outputting various types of data provides comprehensive analysis materials for users, helping them locate faults and evaluate the impacts, which is of great significance for ensuring the safe and stable operation of the power grid and enabling operation and maintenance personnel to quickly master the fault situation and take measures.

[0012] Combined with some embodiments of the first aspect, in some embodiments, the step of inputting the transient process data into a deep reinforcement learning model to obtain the calculation frequency adjustment coefficient of each power grid node specifically includes: using the transient process data as the input of the deep reinforcement learning model, and the deep reinforcement learning model uses the weighted sum of the relative error of the calculation result and the computing resource occupancy rate as the reward function; and outputting the calculation frequency adjustment coefficient of each power grid node.

[0013] By adopting the above technical solution, inputting the transient process data into the deep reinforcement learning model, with the model using the weighted sum of the relative error of the calculation result and the computing resource occupancy rate as the reward function and outputting the calculation frequency adjustment coefficient, can adapt to the calculation requirements of each power grid node, avoid resource waste or insufficient accuracy caused by uniformly calculating the frequency for all nodes, and effectively balance the calculation accuracy and resource consumption.

[0014] In some embodiments in combination with some embodiments of the first aspect, after the step of outputting the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes in response to the user's data export operation, the method further includes: establishing a graphical configuration interface, and displaying the power grid nodes and the power grid connection relationship on the graphical configuration interface; in response to the user's target node operation and simulation parameter input operation on the graphical configuration interface, writing a control script according to the simulation parameters in the target node, where the simulation parameters include the fault type, the fault duration, and the fault impedance, and the control script defines the data transfer rules and the calculation trigger conditions between nodes at each level; and executing the control script according to the power grid connection relationship and the propagation time sequence.

[0015] By adopting the above technical solution, a graphical configuration interface is constructed to display the power grid nodes and the connection relationship, which is convenient for users to operate. After the user inputs the simulation parameters, the platform writes a control script and executes it in combination with the power grid connection relationship and the propagation time sequence. The graphical interface provides an intuitive operation environment, and the user can conveniently set parameters. The script standardizes the calculation process, ensures that data is transferred and calculated according to rules, and enhances the flexibility and controllability of the simulation.

[0016] In some embodiments in combination with some embodiments of the first aspect, after the step of executing the control script according to the power grid connection relationship and the propagation time sequence, the method further includes: according to the execution result of the script, updating the node status information and the calculation progress in the graphical configuration interface in real time.

[0017] By adopting the above technical solution, after the control script is executed, the node status information and the calculation progress in the graphical configuration interface are updated according to the result. The real-time update enables the user to master the simulation dynamics at any time, detect abnormalities in a timely manner, and make adjustments.

[0018] In some embodiments in combination with some embodiments of the first aspect, after the step of outputting the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes in response to the user's data export operation, the method further includes: creating a temporary data buffer, and storing the power grid node data currently being calculated into the temporary data buffer; allocating calculation tasks according to the power grid node hierarchy order, and raising the calculation priorities of the core layer and the near-domain layer to a preset priority threshold.

[0019] By adopting the above technical solution, creating a temporary data buffer can effectively store the power grid node data during calculation, allocate tasks according to the hierarchy, and raise the priorities of the core layer and the near-domain layer. The core layer and the near-domain layer are close to the fault point, and the data changes are crucial, which can accelerate the simulation speed.

[0020] Second aspect, an embodiment of the present application provides an electromagnetic transient real-time simulation platform, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, and the computer program code includes computer instructions. The one or more processors call the computer instructions to cause the electromagnetic transient real-time simulation platform to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0021] Third aspect, an embodiment of the present application provides a computer program product containing instructions. When the computer program product runs on the electromagnetic transient real-time simulation platform, it causes the electromagnetic transient real-time simulation platform to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0022] Fourth aspect, an embodiment of the present application provides a computer-readable storage medium, including instructions. When the instructions run on the electromagnetic transient real-time simulation platform, it causes the electromagnetic transient real-time simulation platform to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0023] It can be understood that the electromagnetic transient real-time simulation platform provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, and will not be elaborated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. In the present application, the power grid structure is presented by constructing a spatio-temporal graph neural network model. The propagation time matrix is obtained by inputting physical parameters, and layers are divided accordingly. The core layer is calculated with a basic time step, and the far domain layer is calculated with a larger time step, reasonably allocating computing resources and reducing consumption. The electrical characteristic parameters are calculated in the core layer to identify the key node set, and the boundary node parameters at its intersection are used as boundary conditions to ensure accurate calculation. Finally, accurate transient process data is obtained, improving the calculation efficiency.

[0025] 2. In the present application, layers are divided according to the propagation time matrix. The core layer is calculated at a high frequency with a basic time step to ensure accuracy, and the far domain layer is calculated at a low frequency with eight times the basic time step to reduce resource consumption. Intermediate layers such as the near domain layer and the transition layer also adopt corresponding time steps according to the propagation time, reducing the calculation amount while ensuring a certain accuracy. Overall, the computing resources are reasonably allocated, the calculation efficiency is improved while ensuring the accuracy of the key area, and the transient process data is efficiently obtained.

[0026] 3. This application constructs a graphical configuration interface to display power grid nodes and connection relationships, facilitating user operation. After the user inputs simulation parameters, the platform writes control scripts, which are executed in combination with the power grid connection relationships and propagation time sequences. The graphical interface provides an intuitive operation environment, enabling users to conveniently set parameters. The scripts standardize the calculation process, ensuring that data is transmitted and calculation triggers are based on rules, enhancing the flexibility and controllability of the simulation. Description of the Drawings

[0027] Figure 1 is a schematic flow diagram of an efficient, autonomous, and controllable electromagnetic transient real-time simulation method in an embodiment of this application; Figure 2 is another schematic flow diagram of an efficient, autonomous, and controllable electromagnetic transient real-time simulation method in an embodiment of this application; Figure 3 is a schematic structural diagram of an entity device of an electromagnetic transient real-time simulation platform in an embodiment of this application. Detailed Embodiments

[0028] The terms used in the following embodiments of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification of this application, the singular forms "a", "an", "the above", "the", and "this" are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations including one or more of the listed items.

[0029] Hereinafter, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as implying or suggesting relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of this application, unless otherwise stated, the meaning of "a plurality" is two or more.

[0030] For ease of understanding, the method provided in this embodiment is described in terms of a process below. Please refer to Figure 1 , which is a schematic flow diagram of an efficient, autonomous, and controllable electromagnetic transient real-time simulation method in an embodiment of this application.

[0031] S101. Construct a spatio-temporal graph neural network model according to power grid devices and the physical connection relationships between the power grid devices. The spatio-temporal graph neural network model constructs the power grid devices as power grid nodes and constructs the physical connection relationships between the power grid devices as network connection relationships.

[0032] Among them, grid equipment refers to physical equipment in a power system such as transformers, circuit breakers, and busbars. The physical connection relationship represents the actual electrical connection method between equipment, such as series connection, parallel connection, etc. The spatio-temporal graph neural network model is a deep learning model that combines temporal and spatial features and is used to describe the dynamic changes of the network structure over time. Grid nodes are the basic units representing equipment in the model. The network connection relationship is the connection method representing the interaction between nodes in the model.

[0033] Specifically, when a digital model of the power grid system needs to be established, first collect basic information such as the type, parameters, and connection method of grid equipment. Then, based on this information, construct a spatio-temporal graph neural network model, map each physical device to a network node with a unique identifier, and establish the network connection between nodes according to the actual connection relationship between the devices. This mapping preserves the physical topology of the power grid and at the same time introduces the learning ability of the graph neural network.

[0034] In some embodiments, the spatio-temporal graph neural network model can be constructed in the following way: Optionally, first establish a device information database containing information such as device type, model, and rated parameters; then assign a unique node ID to each device and create a node object; finally, determine the connection relationship between nodes according to the primary system wiring diagram and establish a connection matrix. Optionally, first initialize the network parameters using a deep learning pre-training model; then import historical operation data for model training; finally, evaluate the model performance through a validation set and optimize the parameters. It can be understood that other network construction methods can also be used, which are not limited here.

[0035] S102. Input the physical parameters of the grid equipment into the feature vector of the network node to obtain the propagation time matrix of the fault signal from the fault point to each grid node.

[0036] Among them, physical parameters refer to parameters describing the electrical characteristics of equipment, such as impedance, capacitance, etc. The feature vector is a multi-dimensional numerical vector used to characterize node characteristics. The propagation time matrix records the time required for the fault signal to propagate between nodes. The fault point refers to the specific location where the fault occurs.

[0037] Specifically, convert the collected physical parameters of the equipment into a standardized numerical form as the component of the feature vector corresponding to the network node. Based on the physical distance and propagation speed between nodes, calculate the time required for the fault signal to propagate from the fault point to each node to form a propagation time matrix. This matrix reflects the propagation characteristics of the fault impact in the system.

[0038] In some embodiments, the propagation time calculation can be achieved in the following ways: Optionally, first establish a parameter conversion model to standardize the physical parameters; then calculate the electrical distance between nodes; finally, solve the time matrix according to the propagation speed. Optionally, first determine the initial fault point; then use the breadth-first search algorithm to calculate the shortest propagation path; finally, generate the time matrix according to the path length and the propagation speed. It can be understood that other parameter processing and time calculation methods can also be used, which are not limited herein.

[0039] S103. Dynamically layer each of the power grid nodes according to the propagation time matrix, divide the power grid nodes into a core layer, a near-domain layer, a transition layer, and a far-domain layer according to the propagation time, and calculate the transient processes of the power grid nodes in each layer with different time steps respectively to obtain the transient process data of each of the power grid nodes.

[0040] Among them, dynamic layering refers to the process of dividing nodes into different levels according to the propagation time. The core layer represents the set of nodes closest to the fault point, the near-domain layer refers to the node area adjacent to the core layer, the transition layer is located in the intermediate area outside the near-domain layer, and the far-domain layer refers to the outermost node area. The time step refers to the time interval between two adjacent calculations in the transient calculation. The transient process data is used to represent the dynamic change process of electrical parameters of the node after the fault occurs, including instantaneous values such as voltage and current.

[0041] After obtaining the propagation time matrix, it is necessary to perform layer processing on the power grid nodes to achieve reasonable allocation of computing resources. Specifically, first set the reference time T as the layering standard, divide the nodes with propagation time less than T into the core layer, those with propagation time between T and 2T into the near-domain layer, those between 2T and 4T into the transition layer, and those greater than 4T into the far-domain layer. Then configure different calculation time steps for each layer. The core layer uses the basic time step △t, the near-domain layer uses 2△t, the transition layer uses 4△t, and the far-domain layer uses 8△t. Through this multi-time-scale calculation strategy, both the calculation accuracy of the key area is ensured and the overall calculation efficiency is improved.

[0042] In some embodiments, node layering and transient calculation can be achieved in various ways: Optionally, first perform clustering analysis on the nodes according to the propagation time to determine the optimal layering threshold; then establish a hierarchical management data structure to record the node information of each layer; finally, configure the calculation task scheduling strategy according to the time steps of each layer. Optionally, first construct a queue-based layering processing model to realize the dynamic promotion and demotion of nodes; then adopt an adaptive time step control algorithm to adjust the calculation frequency according to the change of node state; finally, use a parallel computing framework to improve the calculation efficiency. It can be understood that other layering methods and calculation strategies can also be used to realize the transient process analysis, which are not limited herein.

[0043] S104. Calculate the electrical characteristic parameters of the faulty node in the core layer, identify the set of key nodes, and use the parameters of the inter-layer junction nodes in this set of key nodes as the boundary conditions for the calculation of adjacent layers.

[0044] Among them, the electrical characteristic parameters refer to the key indicators describing the electrical characteristics of the node, such as voltage amplitude, phase angle, frequency, etc. The set of key nodes represents a group of nodes that have an important impact on the system stability. The inter-layer junction node refers to the node that belongs to adjacent layers simultaneously. The boundary conditions are used to determine the boundary characteristics of the calculation area and ensure the continuity and convergence of the calculation.

[0045] After completing the node layering, it is necessary to focus on the calculation accuracy of the core layer and the data transfer between layers. Specifically, first use a high-precision algorithm in the core layer to calculate various electrical parameters of the faulty node, including instantaneous values such as voltage, current, power, etc. Then, based on the change characteristics and topological relationships of these parameters, identify the set of key nodes in the system. For the key nodes located at the inter-layer junction, use their parameters as the known boundary conditions for the calculation of adjacent layers to ensure the continuity and accuracy of the calculation results of each layer.

[0046] In some embodiments, the electrical characteristic calculation and boundary condition processing can be implemented in various ways: Optionally, first establish a refined calculation model for the core layer; then use the sensitivity analysis method to screen key nodes; finally, construct an inter-layer data transfer interface to realize the dynamic update of boundary conditions. Optionally, first use the state estimation algorithm to improve the calculation accuracy of the core layer; then identify the key nodes of the network based on the graph theory algorithm; finally, use the interpolation algorithm to process the inter-layer data transfer. It can be understood that other algorithms and methods can also be used to implement the characteristic parameter calculation and boundary condition processing, which are not limited here.

[0047] S105. Input the transient process data into the deep reinforcement learning model to obtain the calculated frequency adjustment coefficient of each power grid node.

[0048] Among them, the deep reinforcement learning model refers to an artificial intelligence algorithm that combines a deep neural network and reinforcement learning and is used to optimize the system control strategy. The transient process data represents the dynamic response characteristics of the node during the fault, including the change sequence of parameters such as voltage and current over time. The calculated frequency adjustment coefficient refers to the proportional factor used to dynamically adjust the node calculation frequency, and its value range is between 0 and 1. The node calculation frequency represents the time interval for updating the state of the node.

[0049] After obtaining the transient process data of each node, it is necessary to optimize the computing resource allocation to improve the simulation efficiency. Specifically, first, organize and standardize the transient process data according to the time series as the input of the deep reinforcement learning model. The model uses a multi-layer convolutional neural network to extract data features and optimizes the decision-making strategy through a reinforcement learning algorithm. The reward function is composed of the weighted sum of the relative error of the calculation result and the computing resource occupancy rate, which is used to guide the model training. Finally, the computing frequency adjustment coefficient of each node is output to achieve the dynamic optimization allocation of computing resources.

[0050] In some embodiments, the optimization adjustment of the computing frequency can be achieved in various ways: Optionally, first preprocess and extract features from the transient data; then construct a deep reinforcement learning network structure and design the state space and action space; finally, optimize the model parameters through the policy gradient algorithm. Optionally, first establish a data analysis model based on the attention mechanism; then use the double Q-learning algorithm for policy optimization; finally, improve the learning efficiency through experience replay. It can be understood that other deep learning and reinforcement learning methods can also be used to achieve the optimization of the computing frequency, which is not limited here.

[0051] S106. If the computing frequency adjustment coefficient of the target node is greater than the preset threshold, upgrade the target node to the upper-level node for transient calculation, and the computing time step of the upper-level node is smaller than that of the target node.

[0052] Among them, the target node refers to a specific node that needs to adjust the computing frequency. The preset threshold represents the critical value of the computing frequency adjustment coefficient that triggers the node upgrade. The upper-level node refers to a set of nodes with a higher computing level. The computing time step refers to the time interval between two adjacent calculations, and the time step of the upper-level node is smaller to obtain higher computing accuracy.

[0053] After obtaining the computing frequency adjustment coefficient of the node, it is necessary to dynamically adjust the computing level to meet the accuracy requirements. Specifically, first set a preset threshold (such as 0.8) and judge the computing frequency adjustment coefficient of each node. When it is found that the adjustment coefficient of a certain node exceeds the preset threshold, it indicates that the node requires higher computing accuracy. At this time, upgrade it to the upper level for calculation. For example, upgrade the far-domain layer node to the transition layer, and the computing time step is reduced from 8△t to 4△t to improve the computing accuracy. At the same time, update the layer attributes and related connection relationships of the node in the spatio-temporal graph neural network.

[0054] In some embodiments, the dynamic upgrade of nodes can be achieved in various ways: Optionally, first establish a node status monitoring mechanism to detect the calculation frequency adjustment coefficient in real time; then design a hierarchical conversion rule to define the upgrade conditions and processes; finally, reallocate computing resources through a task scheduler. Optionally, first construct a node upgrade manager to maintain the hierarchical information of the nodes; then implement a smooth transition algorithm to ensure calculation continuity; finally, adopt an adaptive threshold adjustment strategy to optimize the upgrade judgment criteria. It can be understood that other methods can also be used to achieve the dynamic hierarchical adjustment of nodes, which is not limited here.

[0055] S107. In response to the user's data export operation, output the transient process data and faults of each power grid node.

[0056] Among them, the data export operation refers to a data extraction command triggered by the user through the system interface. The transient process data represents the dynamic response process of the node during a fault, including the time series of parameters such as voltage, current, and power. The dynamic change data of the fault influence range is used to describe the propagation process of the fault in the system. The parameter alarm information refers to the warning information generated when the monitored parameter exceeds the safety threshold, including the parameter type, the degree of overlimit, and the duration, etc. The data output refers to exporting the calculation results in a specific format for the user to analyze and use.

[0057] After the simulation calculation is completed, it is necessary to output the analysis results according to the user's needs. Specifically, first respond to the user's data export instruction, extract the transient process data of each node from the calculation result database, including the time series of parameters such as voltage, current, and power. Then generate the dynamic expansion process data of the fault influence range based on the propagation time matrix, and record the distribution of the affected nodes at each moment. At the same time, perform threshold checks on the key parameters of each node, and when it is found that the preset safety limit is exceeded, generate alarm information including the specific parameter value, occurrence time, and duration. Finally, organize and output these data in a standard format for subsequent analysis and processing.

[0058] In some embodiments, the export and display of data can be achieved in various ways: Optionally, first establish a data export configuration interface to provide parameter selection and format setting options; then extract and organize the target data from the distributed database; finally, generate an analysis report including charts, numerical values, and alarm information. Optionally, first construct a real-time data visualization platform to support dynamic data display; then implement a multi-dimensional data analysis function to provide fault feature extraction; finally, establish a data export interface to support the export of data in multiple formats. It can be understood that other ways can also be used to achieve data processing and display, which is not limited here.

[0059] The following further describes the more specific process of the method provided in this embodiment. Please refer to Figure 2, which is another process schematic diagram of the high-performance autonomous and controllable electromagnetic transient real-time simulation method in the embodiments of this application.

[0060] S201. Construct a spatio-temporal graph neural network model according to the power grid equipment and the physical connection relationship between the power grid equipment. The spatio-temporal graph neural network model constructs the power grid equipment into power grid nodes and constructs the physical connection relationship between the power grid equipment into network connection relationships.

[0061] It can be understood that this step is similar to step S101 and will not be elaborated here.

[0062] S202. Input the physical parameters of the power grid equipment into the feature vectors of the network nodes to obtain the propagation time matrix of the fault signal from the fault point to each power grid node.

[0063] It can be understood that this step is similar to step S102 and will not be elaborated here.

[0064] S203. Stratify the power grid nodes according to the propagation time matrix. Divide the nodes with propagation time less than T into the core layer and calculate once for each basic time step. Divide the nodes with propagation time between T and 2T into the near-domain layer and calculate once for every two basic time steps. Divide the nodes with propagation time between 2T and 4T into the transition layer and calculate once for every four basic time steps. Divide the nodes with propagation time greater than 4T into the far-domain layer and calculate once for every eight basic time steps to obtain the transient process data of each power grid node.

[0065] The propagation time refers to the physical time required for the fault signal to propagate from the fault point to the specified node. The basic time step is the minimum time interval unit for transient calculation and is used to discretize the continuous physical process. The core layer, near-domain layer, transition layer, and far-domain layer are four calculation regions divided according to the propagation time from near to far. The transient process data includes the transient response data of each node after the fault occurs, such as the dynamic change process of voltage and current.

[0066] The simulation platform adopts a multi-time-scale hierarchical calculation strategy according to the propagation time matrix. Divide the nodes with propagation time less than the characteristic time T into the core layer, and these nodes are calculated once for each basic time step △t; divide the nodes with propagation time between T and 2T into the near-domain layer and calculate once every 2△t; divide the nodes with propagation time between 2T and 4T into the transition layer and calculate once every 4△t; divide the nodes with propagation time greater than 4T into the far-domain layer and calculate once every 8△t. For example, if the basic time step △t is 50 microseconds and the characteristic time T is 1 millisecond, then the core layer nodes are calculated once every 50 microseconds, the near-domain layer is calculated once every 100 microseconds, the transition layer is calculated once every 200 microseconds, and the far-domain layer is calculated once every 400 microseconds, so as to obtain the transient process data of each node.

[0067] S204. Calculate the electrical characteristic parameters of the faulty node in the core layer, identify the set of key nodes, and use the parameters of the inter-layer junction nodes in this set of key nodes as the boundary conditions for the adjacent layer calculation.

[0068] The simulation platform first calculates the electrical characteristic parameters such as voltage, current value, and phase angle at the faulty node in the core layer, and determines the influence range of the fault based on the change rate and amplitude of these parameters. By analyzing the changes in electrical quantities and topological connection relationships of the core layer nodes, a set of nodes that play a key role in the propagation of the transient process is identified. For the junction nodes that belong to two calculation layers at the same time among these key nodes, their state parameters such as voltage and current are used as the known boundary conditions for the adjacent layer node calculation. When specifically implemented, for a system with a rated voltage of 220 kV, when a three-phase short-circuit fault occurs in a certain circuit breaker, first calculate the characteristic parameters of the fault point. For example, the fault current is 20 kA and the voltage drops to 0.1 times the rated value. Identify the circuit breaker, its adjacent circuit breakers, busbars and other equipment as key nodes, and use the voltage and current values of these key nodes for the transient process solution of the adjacent calculation layer.

[0069] S205. Use the transient process data as the input of the deep reinforcement learning model, and the deep reinforcement learning model uses the weighted sum of the relative error of the calculation result and the calculation resource occupancy rate as the reward function.

[0070] The transient process data refers to the time series of state variables of each node in the power grid during the fault process, including the change data of parameters such as voltage and current over time. The deep reinforcement learning model is an artificial intelligence algorithm that combines a deep neural network and reinforcement learning, and optimizes the decision-making strategy by interacting with the environment. The relative error of the calculation result refers to the deviation degree between the simulation result and the theoretical value or reference value. The calculation resource occupancy rate represents the usage level of hardware resources such as the processor and memory. The reward function is a numerical index for evaluating the system performance, and is used to guide the optimization direction of the model.

[0071] The simulation platform inputs the transient process data such as voltage and current of each node into the deep reinforcement learning model according to the time series. The model uses a multi-layer convolutional neural network to extract data features and optimizes the calculation strategy through the reinforcement learning algorithm. The reward function R consists of two parts: the relative error E of the calculation result and the calculation resource occupancy rate U, that is, R = -α×E - β×U, where α and β are weight coefficients. For example, when the relative error of the voltage calculation result of a certain node is 2% and the CPU occupancy rate is 60%, take α = 0.7 and β = 0.3, then the reward value of this node is -0.014 - 0.18 = -0.194, and the model adjusts the calculation strategy according to this reward value.

[0072] S206. Output the calculation frequency adjustment coefficient of each power grid node.

[0073] The calculation frequency adjustment coefficient refers to the proportionality factor used to dynamically adjust the calculation frequencies of each node, and its value range is a real number between 0 and 1. The larger the adjustment coefficient, the higher the calculation frequency of the node needs to be increased; the smaller the adjustment coefficient, the lower the calculation frequency of the node can be decreased. The calculation frequency directly affects the calculation accuracy and calculation resource consumption, and a balance needs to be achieved between ensuring accuracy and efficiency. For example, an adjustment coefficient of 0.8 means that the calculation frequency of the node is increased to 0.8 times the reference frequency of the current layer.

[0074] Based on the training results of the deep reinforcement learning model, the simulation platform generates calculation frequency adjustment coefficients for each power grid node. These adjustment coefficients are calculated through the decision-making network of the optimization model and reflect the required calculation accuracy requirements of the nodes in the current simulation state. In specific implementation, for the system state at a certain moment, if the adjustment coefficient of output node A is 0.9 and that of node B is 0.3, it means that node A requires a higher calculation frequency to ensure accuracy, while node B can appropriately reduce the calculation frequency to save resources. These adjustment coefficients will be used in the subsequent dynamic adjustment process of the calculation frequency.

[0075] S207. If the calculation frequency adjustment coefficient of the target node is greater than the preset threshold, upgrade the target node to the upper-layer node for transient calculation, and the calculation time step of the upper-layer node is smaller than that of the target node.

[0076] The simulation platform checks the calculation frequency adjustment coefficients of each power grid node. When it is found that the adjustment coefficient of a certain node exceeds the preset threshold (such as 0.8), the calculation method of this node is adjusted to the calculation frequency of the upper layer. In specific implementation, if the calculation frequency adjustment coefficient of a 500 kV transformer node in the far field layer is 0.85, exceeding the preset threshold of 0.8, then this node is upgraded from the far field layer (the calculation step is 400 microseconds) to the transition layer (the calculation step is 200 microseconds) for calculation. After the upgrade, the node obtains a higher calculation frequency and a smaller time step, improving the calculation accuracy. At the same time, update the hierarchical attributes and relevant connection relationships of this node in the spatio-temporal graph neural network.

[0077] S208. In response to the user's data export operation, obtain the voltage drop, overcurrent, and power fluctuation parameters of each layer of power grid nodes, and generate parameter warning information according to the preset threshold.

[0078] The data export operation refers to the data extraction command triggered by the user through the simulation platform interface. The voltage drop represents the percentage reduction of the node voltage relative to the rated value. The overcurrent parameter refers to the multiple by which the current value exceeds the rated current. The power fluctuation parameter describes the variation range of active power and reactive power. The parameter alarm information refers to the warning information generated when the monitored parameter exceeds the safety threshold, including the parameter type, the degree of overlimit, and the duration, etc. For example, when the voltage of a certain circuit breaker drops to 70% of the rated value and the current reaches 2.5 times the rated value, an alarm is triggered.

[0079] After receiving the user's data export instruction, the simulation platform extracts the key electrical parameters of the power grid nodes from each calculation layer. For each node, it calculates the voltage drop (percentage relative to the rated voltage), the overcurrent multiple (ratio relative to the rated current), and the power fluctuation value (deviation relative to the steady-state power). When these parameters exceed the preset safety thresholds (such as voltage drop > 30%, overcurrent multiple > 2.0, power fluctuation > 20%), it generates alarm information including the specific parameter values, occurrence time, and duration. Specifically, for the 220kV bus node, when it is detected that the voltage drops to 154kV (voltage drop of 30%), the current reaches 2000A (2.2 times the rated current), and the power fluctuation reaches 25%, it generates an alarm information of "serious reduction of bus voltage, overcurrent, and overlimit of power fluctuation".

[0080] S209. Generate dynamic change data of the fault influence range based on the propagation time matrix.

[0081] The propagation time matrix records the time information of the fault signal propagating from the fault point to each node, and the matrix element represents the time required for the signal to propagate. The fault influence range refers to the power grid area affected by the fault, including the set of nodes where electrical quantities such as voltage and current change significantly. The dynamic change data describes the expansion process of the fault influence range over time, including the number and distribution of affected nodes at each moment. For example, within 100 microseconds after a certain circuit breaker fails, it affects 5 adjacent nodes, and within 200 microseconds, it expands to 15 nodes.

[0082] The simulation platform uses the propagation time matrix to calculate the spatio-temporal expansion process of the fault influence. First, it determines the initial fault point, and then in the order of the propagation time, it marks the affected nodes in sequence. For each time point, it records the number, location, and change value of electrical parameters of the currently affected nodes to form the time-series data of the fault influence range. Specifically, for the line short-circuit fault in the 220kV system, at t = 0, the influence range includes the fault point and the adjacent circuit breaker, at t = 100 microseconds, it expands to the adjacent bus, at t = 200 microseconds, it expands to the transformer, and so on until the propagation of the fault influence is completed. These data reflect the propagation law and influence degree of the fault in the power grid.

[0083] S210. Output the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes.

[0084] The transient process data is a numerical sequence recording the changes of electrical quantities such as voltage, current, and power of power grid nodes over time during a fault, and is stored in a two-dimensional array format of time-numerical values. The dynamic change data of the fault influence range includes the expansion process of the fault influence area over time, and records the spatial distribution and influence degree of the affected nodes at each moment. The parameter warning information is a warning record of electrical parameters exceeding the safety threshold, including information such as parameter type, over-limit value, occurrence time, and duration. The data output uses a standardized data format, such as a CSV file or a binary data stream.

[0085] The simulation platform formats and outputs the three types of calculated data. For the transient process data, organize the parameters such as voltage, current, and power of each node according to the time series to form a data structure of {time, node ID, parameter value}; for the fault influence range data, generate a dynamic record including timestamp, set of affected nodes, and influence degree; for the parameter warning information, output a warning list including node number, warning type, warning level, and occurrence time. Specifically, when implementing for a short-circuit fault in a 220 kV system, the output format is: the transient process data is in the form of {t = 0 ms, Node001, V = 220 kV, I = 1 kA}; the fault influence range data is recorded as {t = 100 μs, [Node001, Node002], Level = 1}; the warning information is recorded as {Node001, overvoltage warning, severe, t = 50 μs}.

[0086] S211. Establish a graphical configuration interface and display the power grid nodes and power grid connection relationships on this graphical configuration interface.

[0087] The graphical configuration interface is a visual operation platform for displaying and operating the power grid model, presented in a two-dimensional or three-dimensional graphical manner. The power grid nodes are represented as different graphical symbols on the interface. For example, a transformer is represented by a "□" symbol, a circuit breaker is represented by a "─┐" symbol, and a busbar is represented by a thick solid line. The power grid connection relationships are represented by connecting lines to indicate the physical connections between devices, and different voltage levels are distinguished by different colors. For example, 500 kV is represented by red, 220 kV is represented by green, and 110 kV is represented by blue.

[0088] The simulation platform constructs a graphical configuration interface based on Web technology and uses the Canvas or SVG technology of HTML5 to draw the power grid topology diagram. The interface is divided into three parts: the main display area, the toolbar, and the parameter configuration area. The main display area draws the power grid nodes and connection relationships, and supports zooming and panning operations; the toolbar provides function buttons such as node selection, connection drawing, and parameter setting; the parameter configuration area displays the detailed parameters of the selected node. Specifically, when implemented, user interaction events are processed through JavaScript. For example, clicking on a node displays parameter information, dragging a node changes its position, and double-clicking on a node opens a detailed configuration panel. The interface updates and displays the operating status of the nodes in real time. For example, normal operation is displayed in green, a fault state is displayed in red, and an alarm state is displayed in yellow.

[0089] S212. In response to the user's target node operation and simulation parameter input operation on the graphical configuration interface, a control script is written in the target node according to the simulation parameters. The simulation parameters include the fault type, fault duration, and fault impedance. The control script defines the data transfer rules and calculation trigger conditions between nodes at each level.

[0090] The simulation parameters include the basic elements defining the fault characteristics: the fault type (such as single-phase grounding, two-phase short circuit, three-phase short circuit, etc.), the fault duration (the time interval from the start to the end of the fault), and the fault impedance (the resistance and reactance values at the fault point). The control script is a program code used to control the simulation process, which defines the data transfer rules (the method and timing of data exchange between nodes) and the calculation trigger conditions (the specific conditions for starting node calculations). The target node operation refers to the interactive operations such as node selection and parameter setting performed by the user on the interface.

[0091] The simulation platform listens for the user's operation events on the graphical interface. When the user selects a target node and inputs simulation parameters, a corresponding control script is automatically generated. Specifically, when the user sets a three-phase short circuit fault on a 220 kV bus node, the fault duration is 100 ms, and the fault impedance is 0.1 + j0.1 Ω, the platform automatically generates a control script similar to "if(t>=0&&t<=100ms){setFault(BusNode, THREE_PHASE, 0.1 + j0.1); calculateNode(CoreLayer, dt = 50 us); transferData(CoreLayer, NearLayer)}". This script defines the fault conditions, the calculation step size of the core layer, and the inter-layer data transfer rules, and is used to guide the subsequent transient simulation calculation process.

[0092] S213. Execute the control script according to the power grid connection relationship and propagation time sequence.

[0093] After step S213, the following steps are further included: According to the execution result of the script, the node status information and calculation progress in the graphical configuration interface are updated in real time.

[0094] The control script is a set of program codes containing calculation logic and control instructions. The grid connection relationship defines the physical connection order and topological structure between nodes. The propagation time order is the node calculation order determined based on the fault signal propagation speed. The node status information includes the operating status of the node (normal, faulty, alarm, etc.) and electrical parameter values (voltage, current, etc.). The calculation progress represents the completion degree of the current simulation calculation, including the number of nodes with completed calculations and the remaining calculation time. For example, the status information of a certain circuit breaker node is "faulty state, voltage 75 kV, current 2000 A", and the calculation progress is "60% completed, estimated remaining 2 minutes".

[0095] The simulation platform executes the control script according to the physical connection order of the grid and the fault propagation time sequence. First, the calculation instructions of the node where the fault point is located are executed, and then, in the order of the spread of the fault impact, the calculation instructions of the connected nodes are executed in sequence. During the execution process, the platform continuously monitors the calculation status of each node. When the node status changes (such as voltage reduction, current increase), the display status of the corresponding node in the graphical interface is immediately updated (such as color change, value update). At the same time, the calculation progress bar and the estimated remaining time are displayed in real time on the interface. Specifically, in the implementation, for the three-phase short-circuit fault simulation of a 220 kV system, the node where the fault point is located is displayed in red and the fault type is marked, and the adjacent nodes are gradually updated in status as the calculation progresses, and the progress bar shows the current calculation completion ratio.

[0096] S214. Create a temporary data buffer area and store the grid node data currently being calculated into this temporary data buffer area.

[0097] The temporary data buffer area is a memory space specifically used to store the intermediate data during the calculation process. The grid node data includes the static parameters of the node (such as rated value, impedance, etc.) and dynamic state variables (such as instantaneous voltage, current value). The calculation status includes information such as the type of calculation in progress, calculation time, and calculation step size. For example, at a certain moment, the data stored in the buffer area includes "node 001, t = 100 μs, voltage 220 kV, current 1000 A, calculation step size 50 μs", etc.

[0098] The simulation platform allocates a temporary data buffer in memory and uses a hash table structure to store the data during the calculation process. The buffer is indexed according to the node ID, and each node's data entry includes a timestamp, status parameters, and calculation control information. When a node enters the calculation state, all its current relevant data is written to the corresponding position in the buffer. In specific implementation, a two-dimensional array with a size of the number of nodes × the number of data items is created as the buffer, and the data items include fields such as {node ID, timestamp, voltage value, current value, power value, calculation state}. For example, for a system with 1000 nodes, and each node stores 10 data items, a 1000×10 array structure is created for temporarily storing the data during the calculation process.

[0099] S215. Allocate the calculation tasks in the order of the power grid node levels, and increase the calculation priorities of the core layer and the near-domain layer to the preset priority threshold.

[0100] The power grid node levels refer to the hierarchical structure of the core layer, near-domain layer, transition layer, and far-domain layer. The calculation task refers to the node status calculation operation that needs to be executed. The calculation priority defines the priority order of task execution, and the higher the value, the higher the priority. The preset priority threshold is the upper limit value of the priority set by the system. For example, in the priority range of 0 - 100, the core layer is set to 90, the near-domain layer is set to 80, the transition layer is set to 60, and the far-domain layer is set to 40.

[0101] The simulation platform uses a multi-threaded task scheduler to allocate the calculation tasks and adopts a priority queue to manage the task execution order. First, all the calculation tasks are grouped according to the node levels they belong to, and the priority of the core layer tasks is set to the highest value (such as 90), and the priority of the near-domain layer tasks is the second highest (such as 80). The scheduler allocates the tasks to the available calculation threads in the order of the highest to the lowest priority. In specific implementation, 4 priority queues are created corresponding to the four levels respectively, the task priorities of the core layer and the near-domain layer are increased to the preset threshold, and a thread pool (such as containing 16 threads) is used to execute the calculation tasks. When the high-priority queue is empty, the tasks in the low-priority queue will be executed. This allocation method ensures that the calculation tasks in the core area and the near-domain area can obtain the calculation resources first, improving the calculation efficiency of the key areas.

[0102] The following describes the electromagnetic transient real-time simulation platform in the embodiments of the present invention application from the perspective of hardware processing. Please refer to Figure 3 , which is a schematic structural diagram of an entity device of the electromagnetic transient real-time simulation platform in the embodiments of the present application.

[0103] It should be noted that Figure 3 The structure of the electromagnetic transient real-time simulation platform shown is only an example, and should not bring any limitations to the functions and usage scopes of the embodiments of the present invention.

[0104] As Figure 3 shown, the electromagnetic transient real-time simulation platform includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage section 308 into the random access memory (RAM) 303, such as executing the methods described in the above embodiments. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0105] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a push button switch, etc.; an output section 307 including a liquid crystal display (LCD), an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read from it can be installed into the storage section 308 as needed.

[0106] Specifically, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication section 309, and / or installed from the removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are executed.

[0107] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present invention, a computer-readable storage medium can be any tangible medium that contains or stores a program, and this program can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0108] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of the present invention. Among them, each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the above-mentioned module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order from that marked in the accompanying drawings.

[0109] Specifically, the electromagnetic transient real-time simulation platform of this embodiment includes a processor and a memory. A computer program is stored on the memory, and when the computer program is executed by the processor, it implements the high-performance, autonomous and controllable electromagnetic transient real-time simulation method provided in the above embodiment.

[0110] On the other hand, the present invention also provides a computer-readable storage medium, which may be included in the electromagnetic transient real-time simulation platform described in the above embodiment; or it may exist separately and not be assembled into the electromagnetic transient real-time simulation platform. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the electromagnetic transient real-time simulation platform, the electromagnetic transient real-time simulation platform is enabled to implement the high-performance, autonomous and controllable electromagnetic transient real-time simulation method provided in the above embodiment.

[0111] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the various embodiments of the present application.

[0112] As used in the foregoing embodiments, depending on the context, the term "when" may be construed to mean "if" or "after" or "in response to determining" or "in response to detecting". Similarly, depending on the context, the phrase "when determining" or "if (the stated condition or event) is detected" may be construed to mean "if determined" or "in response to determining" or "when (the stated condition or event) is detected" or "in response to detecting (the stated condition or event)".

[0113] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the foregoing embodiments can be implemented, and such processes can be completed by relevant hardware instructed by a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the foregoing method embodiments. The foregoing storage media include: various media such as ROM or random access memory RAM, magnetic disks, or optical discs that can store program codes.

Claims

1. An efficient, autonomous and controllable electromagnetic transient real-time simulation method, characterized in that, Applied to an electromagnetic transient real-time simulation platform, the method includes: Construct a spatio-temporal graph neural network model according to power grid equipment and the physical connection relationships between the power grid equipment. The spatio-temporal graph neural network model constructs power grid equipment into power grid nodes and constructs the physical connection relationships between the power grid equipment into network connection relationships; Input the physical parameters of the power grid equipment into the feature vectors of the network nodes to obtain the propagation time matrix of the fault signal from the fault point to each power grid node; Dynamically layer each power grid node according to the propagation time matrix, divide the power grid nodes into a core layer, a near-domain layer, a transition layer, and a far-domain layer according to the propagation time, and calculate the transient processes of the power grid nodes in each layer with different time steps respectively to obtain the transient process data of each power grid node; Calculate the electrical characteristic parameters of the fault node in the core layer, identify the key node set, and use the inter-layer boundary node parameters in the key node set as the boundary conditions for adjacent layer calculations; Input the transient process data into a deep reinforcement learning model to obtain the calculated frequency adjustment coefficient of each power grid node; If the calculated frequency adjustment coefficient of a target node is greater than a preset threshold, upgrade the target node to an upper-layer node for transient calculation, and the calculation time step of the upper-layer node is less than that of the target node; In response to a user's data export operation, output the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes.

2. The method according to claim 1, wherein The step of dynamically layer each power grid node according to the propagation time matrix, divide the power grid nodes into a core layer, a near-domain layer, a transition layer, and a far-domain layer according to the propagation time, and calculate with different time steps respectively specifically includes: Layer the power grid nodes according to the propagation time matrix, divide the nodes with a propagation time less than T into the core layer and calculate once every basic time step, divide the nodes with a propagation time between T and 2T into the near-domain layer and calculate once every two times the basic time step, divide the nodes with a propagation time between 2T and 4T into the transition layer and calculate once every four times the basic time step, divide the nodes with a propagation time greater than 4T into the far-domain layer and calculate once every eight times the basic time step to obtain the transient process data of each power grid node.

3. The method according to claim 1, wherein The step of, in response to a user's data export operation, output the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes specifically includes: In response to a user's data export operation, obtain the voltage drop, overcurrent, and power fluctuation parameters of the power grid nodes in each layer, and generate parameter warning information according to a preset threshold; Generate the dynamic change data of the fault influence range based on the propagation time matrix; Output the transient process data of each power grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes.

4. The method according to claim 1, wherein The step of input the transient process data into a deep reinforcement learning model to obtain the calculated frequency adjustment coefficient of each power grid node specifically includes: Using the transient process data as the input of the deep reinforcement learning model, and using the weighted sum of the calculation result relative error and the calculation resource occupancy rate as the reward function of the deep reinforcement learning model; Output the calculated frequency adjustment coefficient of each grid node.

5. The method according to claim 1, wherein After the step of responding to the user's data export operation and outputting the transient process data of each grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes, the method further includes: Establish a graphical configuration interface and display the grid nodes and grid connection relationships on the graphical configuration interface; In response to the user's target node operation and simulation parameter input operation on the graphical configuration interface, write a control script in the target node according to the simulation parameters, where the simulation parameters include the fault type, fault duration, and fault impedance, and the control script defines the data transfer rules and calculation trigger conditions between nodes at each level; Execute the control script according to the grid connection relationship and the propagation time sequence.

6. The method according to claim 5, characterized in that After the step of executing the control script according to the grid connection relationship and the propagation time sequence, the method further includes: According to the script execution result, update the node status information and calculation progress in the graphical configuration interface in real time.

7. The method according to claim 1, characterized in that, After the step of responding to the user's data export operation and outputting the transient process data of each grid node, the dynamic change data of the fault influence range, and the parameter warning information of the key nodes, the method further includes: Create a temporary data buffer and store the grid node data currently being calculated in the temporary data buffer; Allocate calculation tasks according to the grid node hierarchy order, and increase the calculation priorities of the core layer and the near domain layer to a preset priority threshold.

8. An electromagnetic transient real-time simulation platform, characterized in that, The electromagnetic transient real-time simulation platform includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the electromagnetic transient real-time simulation platform to execute the method according to any one of claims 1-7.

9. A computer-readable storage medium, comprising instructions, characterized in that, When the instruction runs on the electromagnetic transient real-time simulation platform, enable the electromagnetic transient real-time simulation platform to execute the method according to any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product runs on the electromagnetic transient real-time simulation platform, enable the electromagnetic transient real-time simulation platform to execute the method according to any one of claims 1-7.

Images