Power grid transient voltage safety prevention and control method and device

Abstract

The invention provides a power grid transient voltage safety prevention and control method and device, and relates to the technical field of power system analysis. The prevention and control method comprises the following steps: determining an interpretability analysis result of an evaluation result of graph data of a current power grid under an expected fault; if at least one instability node of which the transient voltage safety index does not meet a preset condition exists in the evaluation result of the graph data, generating an adjustment scheme based on the instability node and an interpretability analysis result of the evaluation result; adjusting the node features of the to-be-adjusted nodes of the graph data based on the adjustment scheme until the transient voltage safety indexes of all the nodes in the evaluation result of the graph data meet preset conditions; and determining the adjustment scheme enabling the transient voltage safety indexes of all the nodes to meet the preset conditions as a prevention and control strategy of the current power grid, and performing prevention and control on the current power grid based on the prevention and control strategy. According to the method, interpretability analysis is carried out on the evaluation result, adjustment is carried out according to the interpretability analysis result, and decision support is provided for a prevention and control strategy.

Description

Technical Field

[0001] This disclosure relates to the field of power system analysis technology, specifically to a method and device for preventing and controlling power grid transient voltage. Background Technology

[0002] In recent years, with the increasing penetration rate of new energy sources in the power system, both transmission and distribution networks have exhibited strong volatility and uncertainty, resulting in significant intraday variations in grid operation. Due to the relative scarcity of flexible and adjustable resources in the system, the safety and stability margin of the power system has been continuously compressed, and the operating point is closer to the safety domain boundary than before, making online safety and stability analysis and control of the power grid increasingly important.

[0003] In actual power grid operation, the power grid operation scenario may deviate significantly from the current plan due to factors such as line breaks caused by natural disasters and temporary equipment shutdowns, causing the system to face transient voltage safety risks again. Therefore, it is necessary to carry out online prevention and control research for real-time operation scenarios.

[0004] In existing technologies, preventive control is usually carried out based on transient voltage safety indicators evaluated by models. However, this preventive control method has poor preventive control effect on the power grid. Summary of the Invention

[0005] This disclosure addresses the problems existing in the prior art by providing a method and device for preventing and controlling power grid transient voltage, which can solve the problem of poor prevention and control effect of existing prevention and control methods on the power grid.

[0006] To achieve the above objectives, the technical solution adopted in this disclosure is as follows: A first aspect of this disclosure provides a method for preventing and controlling transient voltage safety in a power grid, comprising: determining the interpretability analysis results of the evaluation results of graphical data of the current power grid under a anticipated fault; the graphical data includes at least one node with node characteristics; the evaluation results include transient voltage safety indicators for each node; if there is at least one unstable node in the evaluation results of the graphical data whose transient voltage safety indicators do not meet preset conditions, then generating an adjustment scheme based on the unstable node and the interpretability analysis results of the evaluation results; the adjustment scheme includes the adjustment direction and adjustment amount of the node characteristics of the node to be adjusted; adjusting the node characteristics of the node to be adjusted in the graphical data based on the adjustment scheme until the transient voltage safety indicators of all nodes in the evaluation results of the graphical data meet the preset conditions; determining the adjustment scheme that makes the transient voltage safety indicators of all nodes meet the preset conditions as the prevention and control strategy for the current power grid, and implementing prevention and control for the current power grid based on the prevention and control strategy.

[0007] In some embodiments of this disclosure, determining the interpretability analysis results of the evaluation results of the graph data of the current power grid under the anticipated fault includes: determining the graph data of the current power grid under the anticipated fault; determining the evaluation results of the graph data of the current power grid under the anticipated fault based on the evaluation module of the pre-built transient voltage security evaluation model and the graph data of the current power grid under the anticipated fault; and determining the interpretability analysis results of the evaluation results of the graph data based on the interpretability analysis module of the transient voltage security evaluation model and the evaluation results of the graph data.

[0008] In some embodiments of this disclosure, the interpretability analysis results include node-level interpretability analysis results and feature-level interpretability analysis results; node features include control variables; the interpretability analysis module based on the transient voltage safety assessment model and the evaluation results of the graph data determine the interpretability analysis results of the graph data evaluation results, including: the interpretability analysis module based on the transient voltage safety assessment model performs node-level interpretability analysis on the graph data evaluation results to obtain node-level interpretability analysis results of the graph data evaluation results; the node-level interpretability analysis results include the attention weight coefficient of each node to itself and the attention weight coefficient of all neighborhood nodes to each node; the interpretability analysis module based on the transient voltage safety assessment model performs feature-level interpretability analysis on the graph data evaluation results to obtain feature-level interpretability analysis results of the graph data evaluation results; the feature-level interpretability analysis results include the quasi-steady-state sensitivity of the transient voltage safety index of each node with respect to any control variable.

[0009] In some embodiments of this disclosure, the formula for calculating the quasi-steady-state sensitivity includes: , In the formula, is the quasi-steady-state sensitivity of the i-th control variable under the anticipated fault c, is the transient voltage safety index after increasing the adjustment amount of the i-th control variable under the anticipated fault c, is the transient voltage safety index after decreasing the adjustment amount of the i-th control variable under the anticipated fault c, and is the adjustment amount of the i-th control variable.

[0010] In some embodiments of this disclosure, an adjustment scheme is generated based on the interpretability analysis results of the unstable node and the evaluation results, including: determining the node-level interpretability analysis results of the unstable node based on the interpretability analysis results of the unstable node and the evaluation results; determining the key influencing nodes of the unstable node based on the node-level interpretability analysis results of the unstable node; the key influencing node is the node with the largest attention weight coefficient for the unstable node; determining the key control variables of the key influencing nodes based on the interpretability analysis results of the key influencing nodes and the evaluation results; the key control variables are the control variables with the largest quasi-steady-state sensitivity; and generating an adjustment scheme based on the key control variables of the key influencing nodes; the adjustment scheme includes the adjustment direction and adjustment amount of the key control variables.

[0011] In some embodiments of this disclosure, the adjustment scheme includes: , In the formula, is the adjustment amount of the i-th control variable. It is the transient voltage safety index constraint threshold, which is the transient voltage safety index after node j control adjustment under the anticipated fault c. It is the fine-grained rule coefficient of the i-th control variable with respect to node j under the expected fault c.

[0012] In some embodiments of this disclosure, determining the graph data of the current power grid under a anticipated fault includes: acquiring the operating data of the current power grid; the operating data includes operating conditions, equipment characteristic parameters, and power grid topology; and generating the graph data of the current power grid under the anticipated fault based on the operating data of the current power grid and the anticipated fault.

[0013] A second aspect of this disclosure provides a power grid transient voltage safety prevention and control device, comprising: a determining unit, configured to determine the interpretability analysis results of the evaluation results of the current power grid under a anticipated fault; the graph data includes at least one node with node characteristics; the evaluation results include the transient voltage safety index of each node; a scheme generation unit, configured to generate an adjustment scheme based on the unstable node and the interpretability analysis results of the evaluation results if there is at least one unstable node in the evaluation results of the graph data whose transient voltage safety index does not meet preset conditions; the adjustment scheme includes the adjustment direction and adjustment amount of the node characteristics of the node to be adjusted; an adjustment unit, configured to adjust the node characteristics of the node to be adjusted in the graph data based on the adjustment scheme until the transient voltage safety index of all nodes in the evaluation results of the graph data meets the preset conditions; and a prevention and control unit, configured to determine the adjustment scheme that makes the transient voltage safety index of all nodes meet the preset conditions as the prevention and control strategy of the current power grid, and implement prevention and control on the current power grid based on the prevention and control strategy.

[0014] This disclosure also provides an electronic device, comprising: a memory for storing at least one instruction; and a processor for calling the instruction stored in the memory to execute the power grid transient voltage safety prevention and control method in the first aspect and any embodiment of the first aspect.

[0015] This disclosure also provides a computer-readable storage medium storing at least one executable instruction, which is loaded and executed by a processor to implement the power grid transient voltage safety prevention and control method in the first aspect and any embodiment of the first aspect.

[0016] This disclosure also provides a computer program product, which includes computer program code. When the computer program code is run by a computer, it causes the computer to execute the power grid transient voltage safety prevention and control method in the first aspect and any embodiment of the first aspect.

[0017] Compared with the prior art, this disclosure has the following beneficial effects: The power grid transient voltage safety prevention and control method provided in this disclosure represents the power grid in the form of graphical data, resulting in more accurate evaluation results. Simultaneously, interpretability analysis is performed on the evaluation results to clearly locate the characteristic contribution of each node in the graphical data, transforming the black-box evaluation into a readable result. Furthermore, the interpretability analysis results are used to generate corresponding adjustment schemes, realizing the conversion from interpretable analysis results to executable adjustment schemes. Finally, the adjustment scheme that makes all node indicators meet the requirements is recorded, and based on this, preventive control of the power grid is carried out, achieving effective preventive control of power grid transient voltage safety. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a power grid transient voltage safety prevention and control method provided in Embodiment 1 of this disclosure; Figure 2 This is a flowchart illustrating a power grid transient voltage safety prevention and control method provided in Embodiment 2 of this disclosure; Figure 3 This is a flowchart illustrating a power grid transient voltage safety prevention and control method provided in Embodiment 3 of this disclosure; Figure 4 This is a schematic diagram of node-level interpretability analysis provided in Embodiment 3 of this disclosure; Figure 5 This is a schematic diagram of a prevention and control strategy provided in an embodiment of this disclosure; Figure 6 This is a schematic diagram of another prevention and control strategy provided in an embodiment of this disclosure; Figure 7 This is a schematic diagram of another prevention and control strategy provided in the embodiments of this disclosure; Figure 8 This is a schematic diagram of another prevention and control strategy provided in the embodiments of this disclosure; Figure 9 This is a schematic diagram of the overall process of a power grid transient voltage safety prevention and control method provided in an embodiment of this disclosure; Figure 10 This is a structural block diagram of a power grid transient voltage safety prevention and control device provided in an embodiment of this disclosure. Detailed Implementation

[0019] The present disclosure will now be further described with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present disclosure and should not be construed as limiting the scope of protection of the present disclosure. It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application.

[0020] The acquisition, transmission, storage, use, and processing of data in this disclosed technical solution comply with relevant national laws and regulations. In the embodiments of this disclosure, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this disclosure, and do not imply that the applicant has already used or necessarily used such solutions.

[0021] All terms used in this disclosure have the same meaning as understood by one of ordinary skill in the art to which this disclosure pertains, unless otherwise specifically defined. It should also be understood that terms defined in general dictionaries should be interpreted as having meanings consistent with their meanings in the context of the relevant art, and not as having idealized or highly formalized meanings, unless expressly defined herein.

[0022] Currently, most transient voltage safety assessments focus on post-fault transient scenarios, using wide-area measurement systems and taking transient time-series data from each node as input. However, for transient voltage safety assessments under anticipated faults, more effective assessment model construction methods are needed, requiring in-depth exploration of the complex relationship between the current steady-state operating conditions and the transient processes of the system under anticipated faults.

[0023] Evaluation methods for relevant data-driven models face performance bottlenecks. Due to the "black box" nature of data-driven models, the evaluation results are often unreliable and have poor preventive and control effects on power grid transient voltage safety.

[0024] Based on this, Embodiment 1 of this disclosure provides a method for preventing and controlling power grid transient voltage, such as... Figure 1 As shown, the specific steps include S101 to S104.

[0025] Step S101: Determine the interpretability analysis results of the evaluation results of the graph data of the current power grid under the expected fault.

[0026] It should be noted that, in the embodiments disclosed herein, the anticipated faults are selected from one or more of a set of anticipated faults, which covers all types of disturbance faults such as short circuits, interruptions, DC blocking, new energy grid disconnection, equipment failure, and lightning strikes / wildfires. In one possible implementation, the set of anticipated faults includes single-phase grounding faults, two-phase grounding faults, three-phase grounding faults, multiple cascading faults, commutation failure faults, circuit breaker malfunctions, transformer inter-turn faults, and high-power impact load step faults.

[0027] In this embodiment of the disclosure, the graph data includes at least one node with node characteristics. In one possible implementation, the graph data includes multiple nodes with node and edge characteristics; in another possible implementation, the graph data includes multiple nodes with node and edge characteristics and their connection relationships to the power grid topology.

[0028] For example, in one specific embodiment, multiple nodes are formed by multiple electrical regions of the current power grid. The node characteristics are composed of electrical quantities such as voltage, frequency, active power, reactive power, and current of the electrical regions of the current power grid. The edge characteristics are composed of the impedance and power magnitude between lines in the electrical regions of the current power grid. After feature normalization, the node characteristics and edge characteristics are arranged in an orderly manner based on the power grid topology connection relationship to form graph data. z 1. Graph data z 1 includes a node feature matrix, an edge feature matrix, and a topological adjacency matrix. The node features include voltage amplitude, frequency offset, injected active power, injected reactive power, and current amplitude. The edge features include the equivalent impedance amplitude and phase angle between the nodes at both ends of the line, the active power currently flowing through the line, the reactive power currently flowing through the line, and the effective value of the current currently flowing through the line. The topological adjacency matrix stores the electrical distance and real-time connectivity status in a weighted form.

[0029] Understandably, graph data z 1. By isomorphically mapping the node feature matrix, edge feature matrix, and topological adjacency matrix, the current operating conditions, power grid topology connections, and anticipated fault scenarios are uniformly embedded into the same high-dimensional feature space, achieving an integrated representation of the operation-topology-fault triple. When the power grid topology changes due to factors such as accidents or temporary maintenance, the corresponding topological adjacency matrix needs to be adjusted accordingly.

[0030] In this embodiment of the disclosure, the evaluation results include the transient voltage safety index of each node. In one possible implementation, the evaluation results include the transient voltage safety index of each node within 0-5 seconds after a fault.

[0031] In this embodiment of the disclosure, the interpretability analysis results include node-level interpretability analysis results, feature-level interpretability analysis results, edge-level interpretability analysis results, critical propagation paths, etc. Among them, node-level interpretability analysis results can provide key node information affecting transient voltage safety for preventive control; feature-level interpretability analysis results can provide key control variable information of related equipment for preventive control; edge-level interpretability analysis results can provide key section information requiring emergency power flow adjustment or switching for preventive control; and critical propagation paths can provide electrical propagation chain information of disturbance energy from the fault source to the unstable node for preventive control.

[0032] Additionally, it should be noted that the purpose of interpretability analysis of the evaluation results is to transform the black-box output of the graph neural network into causal knowledge that the scheduler can understand, so that subsequent adjustments no longer rely on trial and error.

[0033] Step S102: If there is at least one unstable node in the evaluation results of the graph data that does not meet the preset conditions for transient voltage safety index, then an adjustment scheme is generated based on the interpretability analysis results of the unstable node and the evaluation results.

[0034] It should be noted that "not meeting the preset conditions" refers to not meeting the transient voltage safety requirements. In one possible implementation, a transient voltage safety index constraint threshold is set. When the transient voltage safety index of any node exceeds this threshold, the node is determined to not meet the transient voltage safety requirements, i.e., it does not meet the preset conditions, and is identified as an unstable node.

[0035] In this embodiment of the disclosure, the adjustment scheme includes the adjustment direction and adjustment amount of the node features of the node to be adjusted. In one possible implementation, the adjustment scheme includes a set of nodes to be adjusted and a set of features to be adjusted, wherein the set of nodes to be adjusted includes at least one node to be adjusted, and the set of features to be adjusted includes the adjustment direction (increase or decrease) and adjustment amount (per-unit value or dimensional physical quantity) of each feature; in another possible implementation, the adjustment scheme further includes adjusting priority weights.

[0036] The approach to generating adjustment schemes based on unstable nodes and interpretability analysis results is as follows: sort nodes in descending order of absolute contribution value, prioritize reverse correction of the node characteristics with the largest contribution and physical adjustability, and use the fastest trajectory to pull the unstable index back to the qualified range.

[0037] Step S103: Adjust the node characteristics of the nodes to be adjusted in the graph data based on the adjustment scheme until the transient voltage safety index of all nodes in the evaluation results of the graph data meets the preset conditions.

[0038] It should be noted that the node characteristics of the nodes to be adjusted in the embodiments of this disclosure include control variables and / or state variables. In one possible implementation, a small-step iterative approximation method can be used to adjust the node characteristics of the graph data multiple times step by step. After each step adjustment, the evaluation result of the graph data is re-determined until the transient voltage safety index of all nodes in the evaluation result of the graph data meets the preset conditions, at which point the iteration ends.

[0039] This process can converge to a feasible adjustment scheme within seconds, and ensure that the adjustment scheme meets voltage safety requirements without exceeding equipment capacity limits or causing new problems related to thermal stability or transient angle stability.

[0040] In addition, it should be noted that the embodiments disclosed herein adopt a rolling time-domain optimization framework, which updates the sensitivity matrix after each round of evaluation and dynamically corrects the adjustment amount for the next step to avoid the accumulation of linear approximation errors. When the improvement amount of the index is less than the threshold for two consecutive rounds or the number of iterations reaches the upper limit, the process is forcibly terminated and the current optimal adjustment scheme is output to ensure the real-time performance of online applications.

[0041] Step S104: The adjustment scheme that makes the transient voltage safety indicators of all nodes meet the preset conditions is determined as the current power grid prevention and control strategy, and prevention and control are implemented on the current power grid based on the prevention and control strategy.

[0042] In this embodiment of the disclosure, when the transient voltage safety index of all nodes in the evaluation results of the graph data meets the preset conditions, all adjustment schemes that make the transient voltage safety index of all nodes meet the preset conditions are recorded and transformed into the current power grid's preventive control strategy, so that when a fault occurs in the current power grid, preventive control can be implemented on the current power grid based on the preventive control strategy under the corresponding expected fault.

[0043] It should be noted that the prevention and control strategy in this embodiment includes all ultimately converged adjustable points and their characteristic adjustment sets, which can be in the form of JSON structured messages or CIM-E extended files, supporting automatic parsing and conversion into standard control commands by the scheduling master station. The implementation of prevention and control includes a five-step process: dispatcher confirmation, control command issuance, field equipment action, measurement recalibration, and strategy activation flag setting, ensuring that the strategy is transformed from offline calculation to online running point migration.

[0044] In addition, it should be noted that the embodiments of this disclosure can write qualified strategies into the prevention and control strategy library, establish a "predicted fault - strategy" key-value index, and support direct calling when the same type of fault occurs again; at the same time, a strategy validity period and rolling refresh mechanism are set, and the strategy is automatically triggered to recalculate when the power grid topology or operating mode changes significantly, so as to ensure long-term effectiveness.

[0045] The power grid transient voltage safety prevention and control method provided in Embodiment 1 of this disclosure injects anticipated faults into graph data and completes transient voltage safety assessment in a unified feature space. If there are unstable nodes whose assessment results do not meet preset conditions, the interpretability analysis results of the assessment results are used to generate a node feature adjustment scheme with direction and dimension and iterate in a closed loop until the assessment results of all nodes meet the preset conditions. Finally, the converged adjustment amount is solidified into a prevention and control strategy and issued offline to realize the prevention and control of the current power grid.

[0046] Based on Embodiment 1, Embodiment 2 of this disclosure provides a method for preventing and controlling power grid transient voltage, such as... Figure 2As shown, in one possible implementation, the interpretability analysis results of the evaluation results of the graph data of the current power grid under the expected fault are determined, specifically including the following steps S201 to S203.

[0047] Step S201: Determine the graph data of the current power grid under the anticipated fault.

[0048] In one possible implementation, determining the graph data of the current power grid under the anticipated fault may specifically include the following steps S211 to S212.

[0049] Step S211: Obtain the current power grid operation data.

[0050] It should be noted that the operating data includes operating conditions, equipment characteristic parameters, and power grid topology.

[0051] Step S212: Based on the current power grid operation data and anticipated faults, generate graphical data of the current power grid under anticipated faults.

[0052] It should be noted that, in the embodiments of this disclosure, generating graph data of the current power grid under the anticipated fault based on the current power grid operation data and anticipated faults means: based on the current power grid operation data, injecting the fault type, fault location, fault impedance, duration and the state of the topology switch after the fault into the corresponding nodes or edges to generate high-dimensional graph data that simultaneously reflects "operational conditions-power grid topology-fault scenario", so that the subsequent transient voltage safety assessment model can be directly read.

[0053] Additionally, it should be noted that during the fault injection process, only the node characteristics, edge characteristics, and adjacency weights of the fault-related nodes are modified, while the rest of the power grid data remains unchanged.

[0054] Step S202: Based on the evaluation module of the pre-built transient voltage security evaluation model and the graph data of the current power grid under the expected fault, determine the evaluation result of the graph data of the current power grid under the expected fault.

[0055] In this embodiment, a transient voltage safety assessment model needs to be constructed first. Online prevention and control of system transient voltage safety requires a basis on transient voltage safety indicators under different anticipated fault conditions. Therefore, the mathematical model of the transient voltage safety assessment model can be represented by a set of high-dimensional differential-algebraic equations: , In the formula, It is a system state variable; It is a system control variable; and It is the power grid before the fault t 0-Steady-state operating conditions at any given time. Due to the localized nature of reactive power and voltage in the power grid, under different anticipated faults... and power grid topology Under these conditions, the system bus voltage exhibits different transient characteristics due to the differences in dynamic interaction between nearby equipment.

[0056] It should be noted that the transient voltage safety assessment model constructed in this embodiment includes an assessment module. The input of this assessment module is the graph data of the current power grid under the anticipated fault, and the output is the transient voltage safety index of each node of the graph data, i.e., the assessment result of the graph data. In one possible implementation, the assessment module adopts a GNN (Graph Neural Network) model, and the specific construction stages of the GNN model are as follows.

[0057] Phase 1: Offline Data Sample Generation. First, a dynamic model of the power grid of interest is constructed using power system simulation tools. The model data is adjusted based on historical steady-state power grid operation data, and training, testing, and validation sample sets are built based on transient simulations. To cover a sufficiently rich range of system operating scenarios, load levels and generator output in the simulation model data are adjusted within a certain range. To enhance the model's applicability under different power grid topologies and anticipated faults, the power grid topology is adjusted and anticipated faults are set under power grid operating condition constraints. Therefore, the dataset... z It includes combinations of different operating conditions, grid topologies, and anticipated faults. Through transient simulation of the model, quantitative indicators are generated based on the transient voltage curves of each node, and these indicators serve as the true labels for sample evaluation. y .

[0058] Phase Two: Graph Data Construction. The input data structure for the GNN model is graph data, consisting of node features, edge features, and grid topology connections. In the graph data preprocessing stage, information such as steady-state operating conditions, equipment characteristic parameters, and various fault types in the generated sample data needs to be transformed into graph data. Feature selection is based on transient voltage interaction mechanism analysis. After feature normalization, the node features and edge features are arranged in an ordered manner based on the grid topology connections to form the graph data. z 1. When the power grid topology changes due to factors such as accidents or temporary maintenance, the corresponding topology matrix needs to be adjusted accordingly.

[0059] Phase 3: Offline Model Training. In the offline model training phase, a GNN model adapted to the transient voltage problem needs to be constructed first. Then, the training set samples from the graph data obtained in Phase 2 are input into the evaluation model to obtain the corresponding output results. y1. The output of each sample includes the transient voltage safety indicators of each node under the corresponding operating conditions and anticipated faults. The loss is calculated based on the true labels of the samples in the training set and the evaluation results, and then the hyperparameters of the GNN model, such as the transfer function, are updated. Update function and readout function Parameters. During model training, the GNN model with the optimal parameters is calculated and selected based on the validation set.

[0060] Phase Four: Online Model Application. In the online application phase, based on technologies such as power grid state estimation and dynamic parameter identification, the model acquires real-time data required for the current power grid operating conditions and topology. Combined with a set of anticipated faults, preprocessed graph data is generated and fed into the trained GNN model to obtain the transient voltage safety indicators for each node.

[0061] Step S203: Based on the interpretability analysis module of the transient voltage safety assessment model and the assessment results of the graph data, determine the interpretability analysis results of the graph data assessment results.

[0062] It should be noted that the transient voltage safety assessment model in this embodiment of the present disclosure also includes an interpretability analysis module. The input of the interpretability analysis module is the assessment result of the graph data, and the output is the interpretability analysis result of the graph data.

[0063] Interpretability analysis based on model evaluation results can help power grid operators understand the model's internal decision-making mechanism and provide decision support for online prevention and control of the power system. Therefore, this disclosure conducts research at both the node level and the characteristic level based on the proposed transient voltage security assessment model.

[0064] In one possible implementation, the interpretability analysis results include node-level interpretability analysis results and feature-level interpretability analysis results; node features include control variables (such as voltage amplitude, frequency offset, injected active power, injected reactive power, and current amplitude, each represented as a control variable); based on the interpretability analysis module of the transient voltage safety assessment model and the evaluation results of the graph data, the interpretability analysis results of the graph data evaluation results are determined, which may specifically include the following steps S231 and S232.

[0065] Step S231: Based on the transient voltage safety assessment model, the interpretability analysis module performs node-level interpretability analysis on the assessment results of the graph data to obtain the node-level interpretability analysis results of the graph data assessment results.

[0066] It should be noted that the node-level interpretability analysis results include the attention weight coefficient of each node to itself and the attention weight coefficient of all domain nodes to each node.

[0067] In one possible implementation, an attention mechanism is employed for node-level interpretability analysis. At the node level of the transient voltage safety assessment model, attention weight coefficients reflect the relative importance of neighboring nodes to the central node and the central node to itself, helping to clarify key nodes in the model's decision-making. The sum of the attention weight coefficients of all neighboring nodes and the attention weight coefficient of the central node to itself is 1. Physically, a larger coefficient indicates a greater influence of neighboring nodes on the transient voltage recovery of the central node. In the preventative control of this disclosure embodiment, the node-level interpretability analysis results, including attention weight coefficients, can help operators quickly identify the equipment that plays a major role in the poor transient voltage recovery of nodes of interest (i.e., unstable nodes), thereby enabling the implementation of relevant preventative controls.

[0068] The attention weight coefficient can be calculated using the following formula: , In the formula, the attention weight coefficient This indicates the importance of neighboring node j to the center node i in the graph data, while also taking into account the characteristics of the edges between nodes. The impact; These are the learnable parameters of a single-layer feedforward neural network during model training; LeakyReLU is a non-linear activation function. and These are the transformed features of nodes and edges, respectively. It includes the transmission effect of the line on the characteristics of the domain nodes, and is the effect of the domain nodes on... The result of the action.

[0069] Step S232: The interpretability analysis module based on the transient voltage safety assessment model performs feature-level interpretability analysis on the assessment results of the graph data to obtain the feature-level interpretability analysis results of the graph data assessment results.

[0070] It should be noted that the characteristic-level interpretability analysis results include the quasi-steady-state sensitivity of the transient voltage safety index of each node with respect to any control variable.

[0071] Node-level interpretability analysis provides key node information affecting transient voltage safety for preventive control. Building upon this, feature-level interpretability analysis provides key control variable information for related equipment. Sensitivity analysis, by linearizing complex power grid DAEs (Differential-Algebraic Equations), can obtain the quantitative relationships between control variables and state variables at the current operating point of the system, making it a classic tool for power system analysis.

[0072] In one possible implementation, feature-level interpretability analysis can be performed through first-order sensitivity calculation. Under traditional sensitivity analysis methods, the quantitative calculation formula for the interpretability analysis module based on the transient voltage safety assessment model is as follows: , In the formula, and These are system state variables and control variables, respectively, and both belong to the input features of the transient voltage safety assessment model graphical data; Let be the disturbance value of the i-th control variable; It is the sensitivity of the evaluation results of each node with respect to the i-th control variable under the expected fault c of the current power grid operation condition.

[0073] However, traditional static-based sensitivity analysis methods do not consider the impact of control variable disturbances on other state variables in the system, i.e., the quasi-steady-state physical response of the system after control changes. The quasi-steady-state physical response in typical systems includes: generator AVR, the static characteristics of the load with respect to frequency and voltage, and generator automatic speed governors, etc. In practical systems, the control variable... Under disturbance, the system's quasi-steady-state response causes the state variables to change. Changes, i.e. . These are the system state variables before the control variables are disturbed. Therefore, the characteristic-level interpretability analysis based on traditional sensitivity analysis suffers from a decrease in accuracy.

[0074] Therefore, this disclosure introduces quasi-steady-state sensitivity analysis to consider the changes in state variables after the system control variables are disturbed. To capture the quasi-steady-state response, the system input data is updated using the power flow calculation results after the control variables are disturbed.

[0075] For example, in another possible implementation, characteristic-level interpretability analysis can be performed through quasi-steady-state sensitivity calculation. The specific steps and calculation formulas are as follows: 1) For the anticipated fault c, the evaluation module based on the transient voltage safety assessment model obtains the evaluation index under the original input characteristics of the system. , represented as: ; 2) Perturb the characteristics of the control variables of interest, take into account the quasi-steady-state response of the system, and update the model input graph data. , represented as: ; 3) The evaluation module based on the transient voltage safety assessment model re-evaluates the updated graph data to obtain the indicators. , represented as: ; 4) Based on the results of the two assessments, calculate the STVI index for each node with respect to the control variables of interest. Quasi-steady-state sensitivity The calculation formula includes: , In the formula, It is the quasi-steady-state sensitivity of the i-th control variable under the anticipated fault c. It is the transient voltage safety index after increasing the adjustment amount of the i-th control variable under the anticipated fault c. It is the transient voltage safety index after reducing the adjustment amount of the i-th control variable under the anticipated fault c. It is the adjustment amount of the i-th control variable.

[0076] Based on Embodiment 2, Embodiment 3 of this disclosure provides a method for preventing and controlling power grid transient voltage, such as... Figure 3 As shown, in one possible implementation, an adjustment scheme is generated based on the interpretability analysis results of the instability nodes and evaluation results, which may specifically include the following steps S301 to S304: Step S301: Based on the interpretability analysis results of the unstable nodes and the evaluation results, determine the node-level interpretability analysis results of the unstable nodes.

[0077] It should be noted that the node-level interpretability analysis results for determining the unstable node include: calculating the attention weight coefficients of all neighboring nodes and the central node itself towards the unstable node, with the unstable node as the central node. This is to subsequently determine the node with the greatest impact on the transient voltage recovery of the unstable node based on the attention weight coefficients.

[0078] Step S302: Based on the node-level interpretability analysis results of the unstable node, identify the key influencing nodes of the unstable node.

[0079] It should be noted that the key influencing node is the node with the largest attention weight coefficient to the unstable node.

[0080] For ease of explanation, a schematic diagram of node-level interpretability analysis is provided. Based on the current power grid operating conditions, power grid topology, and specific anticipated faults, the attention weight coefficients between nodes are obtained through the interpretability analysis module of the transient voltage security assessment model, and then analyzed in a visual format. Figure 4 As shown, the vertical and horizontal axes represent the feature aggregation side node and the feature propagation side node, respectively. ji Let a represent the attention weight coefficient of node j to node i, which is also the attention coefficient of the neighborhood node to node i. ii This represents the attention weight coefficient of node i towards itself, also known as the self-attention weight coefficient of node i. Figure 4Node 2 is used as an example of an unstable node. All nodes in the domain of node 2 include node 1 and node 5. The attention weight coefficient of node 2 to itself is... The attention weight coefficient of node 1 to node 2 is: The attention weight coefficient of node 5 to node 2 is: Based on the comparison of attention weight coefficients, if This indicates that the associated equipment of node 1 has a major impact on the transient voltage safety assessment of node 2, meaning that node 1 is a key influencing node of node 2.

[0081] Step S303: Based on the interpretability analysis results of key impact nodes and evaluation results, determine the key control variables of key impact nodes.

[0082] It should be noted that, based on the interpretability analysis results of the key impact nodes and assessment results, the key control variables for the key impact nodes are determined as follows: Based on the interpretability analysis results of the key impact nodes and assessment results, the characteristic-level interpretability analysis results for the key impact nodes are determined; based on the characteristic-level interpretability analysis results of the key impact nodes, the key control variables for the key impact nodes are determined. Specifically, determining the key control variables for the key impact nodes based on the characteristic-level interpretability analysis results of the key impact nodes further includes: Based on the characteristic-level interpretability analysis results of the key impact nodes, determining the quasi-steady-state sensitivity of the transient voltage safety index of the key impact node with respect to each control variable. This is to facilitate subsequent determination of the control variable with the greatest impact on the recovery of the transient voltage based on the quasi-steady-state sensitivity.

[0083] Additionally, it should be noted that the key control variable is the one with the highest quasi-steady-state sensitivity.

[0084] Step S304: Generate an adjustment plan based on the key control variables of key influencing nodes.

[0085] Under the influence of factors such as natural disasters or sudden accidents, the power grid operation scenario may deviate from the plan, bringing uncertainty to the safe operation of the system. Therefore, it is necessary to carry out online prevention and control based on offline control decisions.

[0086] In traditional power grid operation, numerical simulation is time-consuming, and a series of operational rules are typically formulated offline based on anticipated fault analysis under typical scenarios to guide online preventive control. However, this approach relies on human experience, has low accuracy, and is difficult to automate. Post-hoc interpretability analysis of transient voltage safety assessment models provides a foundation for constructing refined system rules, effectively supporting online preventive control. Based on the results of model interpretability analysis, the precise quantitative relationship between transient voltage safety indicators and control variables at nodes of interest can be obtained. Due to the locality of reactive power, the control variables that provide key support for the voltage at specific nodes are limited and are often located near the node and the anticipated fault. If all system control variables are selected for analysis, it will introduce information interference to preventive control decisions, and a large number of weakly correlated variables will reduce computational efficiency. Therefore, based on the key node information obtained from the attention coefficient, relevant control variables are selected as the objects of quasi-steady-state sensitivity analysis and refined rule formulation. For nodes with large STVI indicators (such as those exceeding the set threshold requirements), (i.e., unstable nodes), and construct the following refined rules for prevention and control: In the formula, The indicator after adjusting the control of node j under the anticipated fault c. STVI and current benchmark indicators STVI The estimated deviation; It is the i-th control variable in quasi-steady-state sensitivity analysis, such as generator terminal voltage setpoint, VDCOL control parameters, LVRT control parameters, etc. This refers to the refined rule coefficients of the i-th control variable with respect to node j under the anticipated fault c. It should be noted that, in this embodiment, the quasi-steady-state sensitivity of the transient voltage safety index with respect to the i-th control variable under the anticipated fault c is... As a fine-grained rule coefficient. This indicates that controlling variable i is beneficial to improving the transient voltage safety of node j. The larger the absolute value, the more significant the influence of the control variable on the transient voltage.

[0087] It should be noted that the adjustment plan includes the direction and amount of adjustment for key control variables. Considering the magnitude of the fine-grained rule coefficients and the priority order of control variable adjustments, the control variable adjustment plan can be manually formulated or automatically generated. This is to meet the transient voltage constraint. Taking the adjustment of a single control variable as an example, the adjustment scheme may include: , In the formula, is the adjustment amount of the i-th control variable. It is the transient voltage safety index constraint threshold, which is the transient voltage safety index after node j control adjustment under the anticipated fault c. It is the fine-grained rule coefficient of the i-th control variable with respect to node j under the expected fault c.

[0088] Due to the fine rule coefficient This is the result of linearizing the system's current operating point. Considering the strong nonlinearity of transient voltage safety issues, the system needs to be re-evaluated after adjustment to determine if the adjustment meets the requirements; otherwise, it needs to be adjusted again. Finally, the output adjustment result is used for online preventative control. It should be noted that, considering the different types of control variables for each device, standardization can be performed based on parameter adjustment experience and adjustable range to facilitate comparison of sensitivity impact differences. In addition to the above adjustment methods, considering the strong nonlinearity of transient voltage, small step sizes can also be used during automated adjustment. The iterative approximation method involves adjusting step by step, re-evaluating, and updating the fine rule coefficients.

[0089] When a system has multiple anticipated instability faults or multiple unstable nodes, online preventive control needs to consider the coordination of different control variables and prioritize overlapping adjustment directions that have a common effect. For example... Figures 5-8 Taking the two-dimensional space of control variables as an example, this diagram illustrates the coordination of prevention and control decisions under different scenarios.

[0090] Figures 5-8 In the diagram, the black arrow indicates the performance requirements under the anticipated fault condition. The adjustment direction of the corresponding control variables. The red arrow indicates the coordinated decision-making process of the control variables, taking into account the constraints of anticipated faults 1 and 2, with the goal of minimizing the total adjustment amount or adjustment cost.

[0091] exist Figure 5 In the current running point (the node that needs adjustment), the expected fault 1 requirement is met, but the expected fault 2 requirement is not met. To meet the expected fault 2 requirement, according to the fine rule coefficients, increasing control variable 1 has a primary effect, while increasing control variable 2 has a secondary effect. Therefore, in the coordination decision, increasing control variable 1 can be prioritized.

[0092] exist Figure 6 In the initial stage, neither the operating point nor the conditions for anticipated faults 1 and 2 were met. Based on the refined rule coefficients, reducing control variable 1 could satisfy the condition for anticipated fault 2. To satisfy anticipated fault 1, control variable 1 needed to be further reduced. After multiple coordination iterations, the stability requirement was finally met.

[0093] exist Figure 7 In the initial stage, neither the operating point nor the conditions for anticipated faults 1 and 2 were met. According to the refined rule coefficients, increasing control variables 1 and 2 could satisfy anticipated fault 1. However, considering anticipated fault 2, increasing control variable 2 was chosen. Subsequently, to satisfy anticipated fault 2, control variable 1 needed to be further reduced. After multiple coordination iterations, the stability requirement was finally met.

[0094] exist Figure 8 In the process, the operating point meets the requirements of anticipated fault 2, but does not meet the requirements of anticipated fault 1. To meet the requirements of anticipated fault 1, the control variable 1 is increased according to the fine rule coefficient.

[0095] It should be noted that if the adjustment directions of multiple anticipated faults do not overlap, it means that the safety requirements under all anticipated faults cannot be met by coordinating control variables. In this case, operators can prioritize ensuring system safety under high-probability faults based on fault probability or experience, or introduce additional dynamic reactive power sources.

[0096] For ease of understanding, the following description, in conjunction with the illustrations, provides an overall explanation of the prevention and control process of the power grid transient voltage safety prevention and control method provided in this disclosure embodiment.

[0097] like Figure 9 As shown, the overall prevention and control process includes: First, generating graph data based on the current scenario, i.e., generating graph data based on the current power grid operation data combined with anticipated faults; Second, inputting the graph data into the HEGAN model (Heterogeneous Graph Attention Network), and then performing transient voltage safety assessment based on the HEGAN model to obtain the assessment result of the graph data; Third, performing post-event interpretability analysis of the model, i.e., performing interpretability analysis on the assessment result of the graph data to obtain the interpretability analysis result of the graph data assessment result; Fourth, determining whether each node meets the requirements, i.e., judging whether the transient voltage safety indicators of each node meet the requirements based on the assessment result of the graph data. If not, proceed to step five; if the requirements are met, proceed to step eight; Fifth, generating refined prevention and control rules, specifically based on the interpretability analysis of the assessment result of the graph data. The analysis results generate refined prevention and control rules, which are used to instruct the iterative adjustment of key control variables; Step 6: Iterative adjustment of key control variables; Step 7: Update the HEGAN graph data, i.e., update the graph data input to the HEGAN model, and return to Step 1 after the update; Step 8: Determine whether to perform verification. If verification is required, proceed to Step 9; otherwise, proceed to Step 11; Step 9: Perform numerical simulation calculations on the selected scenario; Step 10: Determine whether each node meets the requirements. If the requirements are met, proceed to Step 11; otherwise, return to Step 5; Step 11: Output the prevention and control adjustment strategy.

[0098] For example, in one specific embodiment, the graph data includes three nodes, namely node 1, node 2, and node 3, and the values ​​of each node are obtained based on the HEGAN model. STVI And perform interpretability analysis to obtain the attention coefficient of each node and the relationship between each node. STVI Regarding the quasi-steady-state sensitivity of control variables 1 and 2; determining the sensitivity of each node. STVI Does it meet the requirements? If node 2 at this time... STVI The requirements are not met. Based on the attention coefficients of each node, it is determined that node 1 has the greatest impact on node 2. At the same time, it is determined that... STVI If the quasi-steady-state sensitivity of control variable 1 is maximized, then a fine-grained preventative control rule is generated: adjustment is performed by adjusting control variable 1 at node 1. The fine-grained rule coefficients of control variable 1 are determined by its quasi-steady-state sensitivity, and the adjustment amount of control variable 1 is then determined using these fine-grained rule coefficients. Δx The adjustment plan was obtained; the control variable 1 was changed from... x Become x+Δx Update the feature parameters in the graph data and recalculate the adjusted STVI of node 2. Δx If node 2's STVI Δx If the requirements are met, the output adjustment decision is: adjust control variable 1 of node 1, with the adjustment direction being increase and the adjustment amount being... Δx If node 2's STVI Δx If it still does not meet the requirements, then proceed according to the adjusted STVI. Δx Reconstruct the fine rules, redetermine the adjustment direction and amount, and readjust until the evaluation results of node 2 after adjustment meet the requirements.

[0099] This paper focuses on online voltage prevention and control in real-time operation scenarios. For complex large-scale power grids with high-voltage DC transmission and distributed renewable energy access at the receiving end, a transient voltage safety assessment model is designed. Interpretability analysis is performed using a message-passing graph neural network, providing key decision support for online transient voltage safety prevention and control.

[0100] Based on this, embodiments of this disclosure also provide a power grid transient voltage safety prevention and control device, such as... Figure 10 As shown, the power grid transient voltage safety prevention and control device 100 includes a determination unit 110, a scheme generation unit 120, an adjustment unit 130, and a prevention and control unit 140.

[0101] The determination unit 110 is used to determine the interpretability analysis results of the evaluation results of the graph data of the current power grid under the expected fault; the graph data includes at least one node with node characteristics; the evaluation results include the transient voltage safety index of each node.

[0102] The scheme generation unit 120 is used to generate an adjustment scheme based on the interpretability analysis results of the unstable node and the evaluation results if there is at least one unstable node whose transient voltage safety index does not meet the preset conditions in the evaluation results of the graph data. The adjustment scheme includes the adjustment direction and adjustment amount of the node characteristics of the node to be adjusted.

[0103] The adjustment unit 130 is used to adjust the node characteristics of the nodes to be adjusted in the graph data based on the adjustment scheme, until the transient voltage safety index of all nodes in the evaluation results of the graph data meets the preset conditions.

[0104] The prevention and control unit 140 is used to determine the adjustment scheme that makes the transient voltage safety indicators of all nodes meet the preset conditions as the prevention and control strategy of the current power grid, and to implement prevention and control on the current power grid based on the prevention and control strategy.

[0105] For specific details and benefits of the power grid transient voltage safety prevention and control device provided in the embodiments of this disclosure, please refer to the above description of the power grid transient voltage safety prevention and control method, which will not be repeated here.

[0106] This disclosure also provides an electronic device, comprising: a memory for storing at least one instruction; and a processor for calling the instruction stored in the memory to execute the power grid transient voltage safety prevention and control method in any of the above embodiments.

[0107] This disclosure also provides a computer-readable storage medium storing at least one executable instruction, which is loaded and executed by a processor to implement the power grid transient voltage safety prevention and control method in any of the above embodiments.

[0108] This disclosure also provides a computer program product, which includes computer program code. When the computer program code is run by a computer, it causes the computer to execute the power grid transient voltage safety prevention and control method in any of the above embodiments.

[0109] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0110] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0111] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0113] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0114] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0115] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0116] It should be noted that the terms "first," "second," and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different parts. Terms such as "including" or "contains" mean that the element preceding the word covers the element listed after the word, and do not exclude the possibility of covering other elements as well.

[0117] Although operations are described in a specific order in the accompanying drawings in this disclosure, it should not be construed as requiring these operations to be performed in the specific order or serial order shown, or requiring all of the shown operations to obtain the desired result. In certain environments, multitasking and parallel processing may be advantageous.

[0118] Finally, it should be noted that the above content is only used to illustrate the technical solution of this disclosure, and is not intended to limit the scope of protection of this disclosure. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of this disclosure do not depart from the substance and scope of the technical solution of this disclosure.

Claims

1. A method for preventing and controlling transient voltage in power grids, characterized in that, include: Determine the interpretability analysis results of the evaluation results of the current power grid under the anticipated fault conditions; The graph data includes at least one node with node characteristics; the evaluation results include the transient voltage safety index of each node. If at least one unstable node fails to meet the preset conditions for transient voltage safety index in the evaluation results of the graph data, an adjustment scheme is generated based on the unstable node and the interpretability analysis results of the evaluation results; the adjustment scheme includes the adjustment direction and adjustment amount of the node characteristics of the node to be adjusted. Based on the adjustment scheme, the node characteristics of the nodes to be adjusted in the graph data are adjusted until the transient voltage safety index of all nodes in the evaluation results of the graph data meets the preset conditions. The adjustment scheme that ensures the transient voltage safety indicators of all nodes meet the preset conditions is determined as the preventive control strategy for the current power grid, and preventive control is implemented on the current power grid based on the preventive control strategy.

2. The power grid transient voltage safety prevention and control method as described in claim 1, characterized in that, The interpretability analysis results of the evaluation results of the graph data for determining the current power grid under the anticipated fault include: Determine the graph data of the current power grid under the anticipated fault; Based on the evaluation module of the pre-built transient voltage security evaluation model and the graph data of the current power grid under the expected fault, the evaluation result of the graph data of the current power grid under the expected fault is determined. Based on the interpretability analysis module of the transient voltage safety assessment model and the assessment results of the graph data, the interpretability analysis results of the assessment results of the graph data are determined.

3. The power grid transient voltage safety prevention and control method as described in claim 2, characterized in that, The interpretability analysis results include node-level interpretability analysis results and feature-level interpretability analysis results; the node features include control variables; the interpretability analysis module based on the transient voltage safety assessment model and the assessment results of the graph data determines the interpretability analysis results of the graph data assessment results, including: Based on the interpretable analysis module of the transient voltage safety assessment model, node-level interpretability analysis is performed on the assessment results of the graph data to obtain the node-level interpretability analysis results of the graph data assessment results; the node-level interpretability analysis results include the attention weight coefficient of each node to itself and the attention weight coefficient of all neighborhood nodes to each node; Based on the interpretable analysis module of the transient voltage safety assessment model, feature-level interpretability analysis is performed on the assessment results of the graph data to obtain the feature-level interpretability analysis results of the assessment results of the graph data; the feature-level interpretability analysis results include the quasi-steady-state sensitivity of the transient voltage safety index of each node with respect to any control variable.

4. The power grid transient voltage safety prevention and control method as described in claim 3, characterized in that, The formula for calculating the quasi-steady-state sensitivity includes: ， In the formula, is the quasi-steady-state sensitivity of the i-th control variable under the anticipated fault c, is the transient voltage safety index after increasing the adjustment amount of the i-th control variable under the anticipated fault c, is the transient voltage safety index after decreasing the adjustment amount of the i-th control variable under the anticipated fault c, and is the adjustment amount of the i-th control variable.

5. The power grid transient voltage safety prevention and control method as described in claim 3, characterized in that, The adjustment plan is generated based on the interpretability analysis results of the unstable node and the evaluation results, including: Based on the interpretability analysis results of the unstable node and the evaluation results, the node-level interpretability analysis results of the unstable node are determined; Based on the node-level interpretability analysis results of the unstable node, the key influencing nodes of the unstable node are identified; the key influencing nodes are the nodes with the largest attention weight coefficients for the unstable node. Based on the interpretability analysis results of the key impact nodes and the evaluation results, the key control variables of the key impact nodes are determined; the key control variables are the control variables with the highest quasi-steady-state sensitivity. Based on the key control variables of the key influencing nodes, the adjustment plan is generated; the adjustment plan includes the adjustment direction and adjustment amount of the key control variables.

6. The power grid transient voltage safety prevention and control method according to any one of claims 3-5, characterized in that, The adjustment plan includes: ， In the formula, is the adjustment amount of the i-th control variable. It is the transient voltage safety index constraint threshold, which is the transient voltage safety index after node j control adjustment under the anticipated fault c. It is the fine-grained rule coefficient of the i-th control variable with respect to node j under the expected fault c, that is, the quasi-steady-state sensitivity of the i-th control variable with respect to node j under the expected fault c.

7. The power grid transient voltage safety prevention and control method as described in claim 2, characterized in that, The determination of the graph data of the current power grid under the anticipated fault includes: Obtain the current power grid's operating data; the operating data includes operating conditions, equipment characteristic parameters, and power grid topology. Based on the current power grid's operating data and anticipated faults, graphical data of the current power grid under anticipated faults is generated.

8. A power grid transient voltage safety prevention and control device, characterized in that, include: The determination unit is used to determine the interpretability analysis results of the evaluation results of the graph data of the current power grid under the anticipated fault. The graph data includes at least one node with node characteristics; the evaluation results include the transient voltage safety index of each node. The scheme generation unit is used to generate an adjustment scheme based on the instability node and the interpretability analysis results of the evaluation results if there is at least one unstable node in the evaluation results of the graph data whose transient voltage safety index does not meet the preset conditions. The adjustment scheme includes the adjustment direction and adjustment amount of the node characteristics of the node to be adjusted; An adjustment unit is used to adjust the node characteristics of the nodes to be adjusted in the graph data based on the adjustment scheme, until the transient voltage safety index of all nodes in the evaluation result of the graph data meets the preset conditions. The prevention and control unit is used to determine the adjustment scheme that makes the transient voltage safety indicators of all nodes meet the preset conditions as the prevention and control strategy of the current power grid, and to implement prevention and control on the current power grid based on the prevention and control strategy.

9. An electronic device, characterized in that, include: Memory, used to store at least one instruction; as well as A processor is configured to invoke instructions stored in the memory to execute the power grid transient voltage safety prevention and control method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one executable instruction, which is loaded and executed by a processor to implement the power grid transient voltage safety prevention and control method as described in any one of claims 1-7.

Images