A smart substation safety measure checking method and system

Abstract

The application relates to the technical field of intelligent substations, and discloses an intelligent substation safety measure checking method and system. The method comprises the following steps: collecting multi-source real-time monitoring data of a target intelligent substation, and performing edge federal aggregation on the multi-source real-time monitoring data to obtain collaborative feature data; constructing a dynamic topology graph of the target intelligent substation based on the collaborative feature data, and performing short-term and temporary fault prediction on the target intelligent substation to obtain a short-term and temporary fault prediction result; performing probability reasoning on each short-term and temporary fault event according to a fault probability propagation model to obtain a cascading failure posterior probability of the corresponding short-term and temporary fault event; performing multi-objective optimization sorting on each safety measure strategy to obtain a safety measure priority sequence; and performing simulation verification on the safety measure priority sequence according to a digital twin to obtain a checking report. The application forms a complete closed loop from risk analysis, rapid verification to strategy optimization, and provides efficient and intelligent technical support for safe and stable operation of the intelligent substation.

Description

Technical Field

[0001] The present invention relates to the technical field of smart substations, and in particular to a method and system for verifying installation measures of smart substations. Background Art

[0002] Smart substations, as the core hubs of power systems, are rapidly increasing in automation. Secondary system safety verification, a key component in ensuring the safe and stable operation of smart substations, is crucial for ensuring reliable power supply and preventing power accidents. However, current secondary system safety verification technology faces challenges in practical application, hindering the safe and efficient operation of smart substations.

[0003] In terms of dynamic adaptability, traditional static calibration methods perform calculations and analysis based on fixed grid topology and operating parameters. They are unable to adapt to dynamic changes in grid topology due to the integration of new energy sources or load fluctuations, resulting in delayed calibration results. In smart substations, key data such as equipment status data, topology information, and protection settings are stored in different systems. There is a lack of effective data interaction and sharing mechanisms between systems, resulting in data silos and difficulties in integrating multi-source data. Existing calibration methods have low fault tolerance and lack sufficient robustness in the face of abnormal scenarios such as communication interruptions and protection misoperation. In these abnormal situations, existing methods are prone to misjudgment, which in turn leads to unnecessary misoperation or ineffective prevention of power accidents. In complex scenarios such as multi-circuit parallel operation, existing calibration technologies have difficulty automatically and accurately analyzing and processing these complex situations, often requiring professional technicians to make manual judgments and decisions based on their extensive experience. This verification method, which relies on manual experience, is not only inefficient and unable to meet the real-time operation and rapid decision-making requirements of smart substations, but is also susceptible to human factors, leading to misjudgments and potential risks to the safe operation of the power grid. As smart substations continue to expand in scale and their operational scenarios become increasingly complex, the number of possible fault combinations is growing exponentially. Rule-based verification methods require pre-defined verification rules for each possible fault combination, making the verification process time-consuming.

[0004] Therefore, an innovative safety measure verification method is urgently needed to meet the needs of rapid development and safe and stable operation of smart substations. Summary of the Invention

[0005] In order to solve the problems of insufficient dynamic adaptability, difficulty in multi-source data integration, low fault tolerance, strong manual dependence and verification efficiency bottleneck in the existing smart substation secondary system safety measure verification technology, the present invention provides a smart substation safety measure verification method and system.

[0006] In a first aspect, an embodiment of the present invention provides a method for verifying security measures in a smart substation, comprising:

[0007] Collect multi-source real-time monitoring data of the target smart substation, and perform edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data;

[0008] Constructing a dynamic topology map of the target smart substation based on the collaborative feature data, and performing short-term fault prediction on the target smart substation to obtain a short-term fault prediction result, wherein the short-term fault prediction result includes a number of short-term fault events and a prediction probability corresponding to each of the short-term fault events;

[0009] Constructing a fault probability propagation model of the target smart substation based on the prior knowledge system and the collaborative feature data, and performing probabilistic reasoning on each of the short-term impending fault events according to the fault probability propagation model to obtain a posterior probability of a cascading failure corresponding to the short-term impending fault event;

[0010] Deducing and analyzing the dynamic topology map and the posterior probability of cascading failures to obtain a number of safety measures, and performing multi-objective optimization sorting on each of the safety measures to obtain a safety measure priority sequence;

[0011] A digital twin of the target smart substation is constructed based on the digital mirror data of the power grid equipment, and the safety measure priority sequence is simulated and verified according to the digital twin to obtain a verification report, wherein the verification report includes a determination result of whether the operating status of the target smart substation meets the preset safety criteria after the execution of each safety measure strategy.

[0012] Preferably, after constructing the digital twin of the target smart substation based on the digital mirror data of the power grid equipment and simulating and verifying the security measure priority sequence according to the digital twin to obtain a verification report, the method further includes:

[0013] Performing a dual-loop optimization on the fault probability propagation model and the digital twin based on the verification report to obtain a dual-loop feedback result, wherein the dual-loop optimization includes an inner loop updating the fault probability propagation model parameters and an outer loop correcting the initial state of the digital twin;

[0014] Each of the security measures strategies is optimized based on the dual-loop feedback results to obtain a corresponding dynamic security measure strategy.

[0015] Preferably, the collecting of multi-source real-time monitoring data of the target smart substation and performing edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data includes:

[0016] Collect multi-source real-time monitoring data of the target smart substation through several edge nodes deployed in the target smart substation, wherein the multi-source real-time monitoring data includes SCADA telemetry data, PMU synchrophasor data, protection device status signals, and equipment sensor data;

[0017] Preprocessing the multi-source real-time monitoring data based on each edge node to obtain local feature data corresponding to the edge node, wherein the preprocessing includes outlier removal and timestamp alignment;

[0018] Each of the local feature data is aggregated based on a federated learning framework to obtain collaborative feature data.

[0019] Preferably, the constructing of a dynamic topology map of the target smart substation based on the collaborative feature data, and performing short-term fault prediction on the target smart substation to obtain a short-term fault prediction result includes:

[0020] Constructing a topology map of the target smart substation based on a graph neural network, where the nodes of the topology map represent devices, and the edge weights of the topology map represent the connection reliability between devices;

[0021] Updating the topology map based on the collaborative feature data to obtain a dynamic topology map;

[0022] The time series status data of the target smart substation is obtained, and short-term fault prediction is performed on the time series status data based on a long short-term memory network to obtain a short-term fault prediction result.

[0023] Preferably, the constructing of the fault probability propagation model of the target smart substation based on the prior knowledge system and the collaborative feature data, and performing probabilistic reasoning on each of the short-term fault events according to the fault probability propagation model to obtain the posterior probability of a cascading failure corresponding to the short-term fault event, includes:

[0024] Based on the prior knowledge system and the collaborative feature data, a fault probability propagation model of the target smart substation is established using a Bayesian network as a framework, wherein the prior knowledge system includes a historical fault tree analysis and an expert rule base, the nodes of the fault probability propagation model represent fault events, and the directed edges of the fault probability propagation model represent causal relationships between fault events;

[0025] Based on the fault probability propagation model, a variational inference algorithm is used to perform probability inference on each of the short-term impending fault events to obtain a posterior probability of a cascading failure corresponding to the short-term impending fault event.

[0026] Preferably, the deduction and analysis of the dynamic topology map and the posterior probability of cascading failures to obtain a number of safety measures, and the multi-objective optimization sorting of each of the safety measures to obtain a safety measure priority sequence, include:

[0027] Deducing and analyzing the dynamic topology map and the posterior probability of cascading failures according to different risk scenarios to obtain a safety strategy for each risk scenario;

[0028] Using the TOPSIS algorithm to perform multi-objective optimization on each of the security measures strategies to determine the priority of the corresponding security measures strategies;

[0029] Each of the security measures strategies is sorted based on the priority to obtain a security measure priority sequence.

[0030] Preferably, the step of constructing a digital twin of the target smart substation based on the digital mirror data of the power grid equipment, and performing simulation verification on the security measure priority sequence according to the digital twin to obtain a verification report includes:

[0031] Using digital twin technology to model and map the digital mirror data of power grid equipment in real time to obtain a digital twin of the target smart substation;

[0032] Simulating and executing the security measure priority sequence based on the digital twin to obtain dynamic characteristic simulation data;

[0033] The dynamic characteristic simulation data is verified for compliance based on preset safety criteria to obtain a verification report.

[0034] Preferably, performing dual-loop optimization on the fault probability propagation model and the digital twin based on the verification report to obtain a dual-loop feedback result includes:

[0035] performing inner-loop optimization on the parameters of the fault probability propagation model based on the verification report to obtain dynamic fault probability propagation model parameters;

[0036] Performing outer-loop correction on the initial state of the digital twin based on the verification report to obtain a corrected initial state of the digital twin;

[0037] Based on the dynamic fault probability propagation model parameters and the modified digital twin initial state, a dual-loop feedback result is constructed.

[0038] Preferably, the optimizing each of the security measures based on the dual-loop feedback results to obtain a corresponding dynamic security measure strategy includes:

[0039] Based on the dual-loop feedback results, each of the security measures strategies is subjected to reinforcement learning iterative optimization, and the verification accuracy and verification response time are used as reward functions to obtain the corresponding dynamic security measures strategy.

[0040] In a second aspect, an embodiment of the present invention provides a smart substation security verification system, including:

[0041] A multi-source data acquisition and fusion module is used to collect multi-source real-time monitoring data of the target smart substation and perform edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data;

[0042] a dynamic topology perception and fault prediction module, configured to construct a dynamic topology map of the target smart substation based on the collaborative feature data, and perform short-term fault prediction on the target smart substation to obtain a short-term fault prediction result, wherein the short-term fault prediction result includes a number of short-term fault events and a predicted probability corresponding to each of the short-term fault events;

[0043] a fault probability propagation analysis module, configured to construct a fault probability propagation model for the target smart substation based on a priori knowledge system and the collaborative feature data, and to perform probabilistic reasoning on each of the short-term impending fault events according to the fault probability propagation model to obtain a posterior probability of a cascading failure corresponding to the short-term impending fault event;

[0044] a safety measure generation module, configured to deduce and analyze the dynamic topology map and the posterior probability of cascading failures to obtain a plurality of safety measure strategies, and perform multi-objective optimization sorting on each of the safety measure strategies to obtain a safety measure priority sequence;

[0045] A simulation verification module is used to construct a digital twin of the target smart substation based on the digital mirror data of the power grid equipment, and simulate and verify the safety measure priority sequence according to the digital twin to obtain a verification report, wherein the verification report includes a judgment result on whether the operating status of the target smart substation meets the preset safety criteria after the execution of each safety measure strategy.

[0046] Compared with the existing technology, the embodiment of the present invention provides a method and system for verifying safety measures for smart substations. Its beneficial effects are as follows: based on the edge federation aggregation and fault probability propagation model, it effectively integrates multi-source heterogeneous data and quantifies the risk of cascading failures, breaking through the bottleneck of traditional methods in dealing with uncertainty problems; relying on the digital mirror of power grid equipment and the optimized digital twin, it compresses the simulation delay from minutes to milliseconds, realizing real-time verification of safety measures strategies; introducing a reinforcement learning mechanism to continuously optimize safety measures strategies and accurately adapt to new power system operation modes. The various links of the present invention work together to form a complete closed loop from risk analysis, rapid verification to strategy optimization, providing efficient and intelligent technical guarantees for the safe and stable operation of smart substations. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flow chart of a method for verifying security measures in a smart substation according to an embodiment of the present invention;

[0048] Figure 2 This is another flowchart of a method for verifying security measures in a smart substation according to an embodiment of the present invention;

[0049] Figure 3 This is a schematic structural diagram of a smart substation safety measures verification system according to an embodiment of the present invention;

[0050] Reference numerals:

[0051] 1. Multi-source data acquisition and fusion module; 2. Dynamic topology perception and fault prediction module; 3. Fault probability propagation analysis module; 4. Safeguard generation module; 5. Simulation verification module. DETAILED DESCRIPTION

[0052] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0053] In describing the present invention, it should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in the present invention in specific circumstances.

[0054] like Figure 1 As shown, it is a flow chart of a method for verifying the installation measures of a smart substation according to an embodiment of the present invention. The embodiment of the present invention provides a method for verifying the installation measures of a smart substation, comprising the steps of:

[0055] S1. Collect multi-source real-time monitoring data of the target smart substation and perform edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data;

[0056] Specifically, step S1 includes:

[0057] 1) Collect multi-source real-time monitoring data of the target smart substation through several edge nodes deployed at the target smart substation;

[0058] Multi-source real-time monitoring data includes SCADA telemetry data, PMU synchronized phasor data, protection device status signals and equipment sensor data.

[0059] The target smart substation in this embodiment is a 220kV substation. Six edge nodes are deployed in the 220kV substation. Each node is equipped with NVIDIA Jetson AGX Xavier and transmits data through the Kafka message queue.

[0060] The following is a detailed description of the method for collecting multi-source real-time monitoring data in this embodiment:

[0061] a) SCADA telemetry data: The SCADA system receives telemetry data such as circuit breaker open / close position, busbar voltage, and line current through the IEC 61850 protocol, with a sampling rate of 1 Hz and a data granularity of minutes.

[0062] b) PMU synchronized phasor data: PMU (synchronized phasor measurement unit) collects three-phase voltage / current phasors at a high frequency of 120Hz (with an accuracy of 0.1%), with a timestamp synchronization error of no more than 1 microsecond, supporting wide-area measurement;

[0063] c) Protection device status signal: The protection action signal (distance protection trip command, differential protection start signal) is obtained in real time through the GOOSE protocol, with a transmission delay of no more than 2 milliseconds;

[0064] d) Device sensor data: IoT data such as device temperature (accuracy error does not exceed ±0.5°C), vibration frequency (10Hz~100Hz), and humidity are collected via LoRaWAN with a sampling rate of 10Hz.

[0065] 2) Preprocess multi-source real-time monitoring data based on each edge node to obtain local feature data of the corresponding edge node;

[0066] Specifically, each edge node is responsible for preprocessing multi-source real-time monitoring data within its jurisdiction to obtain local feature data for that edge node. Preprocessing includes outlier removal and timestamp alignment. In this embodiment, outlier removal involves identifying outliers in SCADA telemetry data using a Grubbs test (with a significance level of 0.01), triggering an alarm when a sudden voltage change exceeds ±10% of the rated value. Timestamp alignment involves linearly interpolating missing data points using a sliding window interpolation method.

[0067] Furthermore, the preprocessing also includes feature extraction. Each edge node extracts features from its own multi-source real-time monitoring data that has undergone outlier removal and timestamp alignment to obtain local feature data of the corresponding edge node.

[0068] 3) Aggregate each local feature data based on the federated learning framework to obtain collaborative feature data.

[0069] To address the data leakage issue caused by traditional data fusion methods, which directly upload raw data to the cloud for fusion, this step proposes a federated learning solution: "data remains static, model moves." Each edge node retains only local feature data and transmits gradient parameters to the cloud through encrypted transmission. The cloud aggregates these gradient parameters, updates the global model, and then sends them back to the edge nodes, forming collaborative feature data for global model training. Specifically, this embodiment uses the federated learning framework PySyft to aggregate gradient parameters across edge nodes to protect data privacy.

[0070] It should be noted that collaborative feature data is a standardized feature set formed by cross-system collaborative processing of multi-source heterogeneous data under the edge computing and federated learning framework. Its core value lies in solving the problems of data silos and privacy protection in smart substations.

[0071] S2. Construct a dynamic topology map of the target smart substation based on the collaborative feature data, and perform short-term fault prediction on the target smart substation to obtain a short-term fault prediction result;

[0072] Specifically, step S2 includes:

[0073] 1) Construct a topological map of the target smart substation based on a graph neural network;

[0074] The nodes in the topology map represent devices. The status of each device consists of its type and operating status. The edge weights in the topology map indicate the reliability of the connection between devices. For example, when a circuit breaker is closed, the corresponding edge weight increases, indicating a reliable connection; while when a transformer is under maintenance, the corresponding edge weight decreases, indicating a possible connection failure.

[0075] 2) Update the topology map based on collaborative feature data to obtain a dynamic topology map;

[0076] The collaborative feature data is input and the topology map is updated through the message passing mechanism of the graph neural network to obtain a dynamic topology map. In one embodiment, when a change in the circuit breaker state is detected, the shortest path is recalculated using the Dijkstra algorithm and the device connectivity matrix is ​​updated.

[0077] 3) Obtain the time series status data of the target smart substation, and perform short-term fault prediction on the time series status data based on the long short-term memory network to obtain the short-term fault prediction result.

[0078] It should be noted that before using the LSTM network for short-term fault prediction, it already uses historical fault data to learn time series features. The short-term fault prediction results include several short-term fault events and the predicted probability corresponding to each short-term fault event. In one embodiment, the target smart substation's circuit breaker operation records and load changes (with a 5-minute time step) over the past hour are collected as time series state data. Short-term fault prediction is performed on this time series state data using the LSTM network, resulting in an overload probability for the next 5 minutes.

[0079] S3. Build a fault probability propagation model for the target smart substation based on the prior knowledge system and collaborative feature data. Perform probabilistic reasoning on each short-term fault event according to the fault probability propagation model to obtain the posterior probability of cascading failures corresponding to the short-term fault event.

[0080] Specifically, step S3 includes:

[0081] 1) Based on the prior knowledge system and collaborative feature data, a fault probability propagation model for the target smart substation is established using a Bayesian network architecture;

[0082] The prior knowledge system is used to construct the network topology, and the collaborative feature data is used as the real-time evidence input node to form a Bayesian reasoning framework of "prior structure + a posteriori evidence", thereby obtaining the fault probability propagation model of the target smart substation.

[0083] Specifically, the prior knowledge system includes historical fault tree analysis and expert rule base. The nodes of the fault probability propagation model represent fault events, and the directed edges of the fault probability propagation model represent the causal relationship between fault events.

[0084] Furthermore, in this embodiment, when constructing a fault probability propagation model based on a Bayesian network architecture, 12 typical fault events are used as nodes. Fault events include but are not limited to busbar differential protection misoperation (PM) and secondary circuit disconnection (SD). A conditional probability table is defined for each node, and directed edges are established through historical fault tree analysis and the expert rule base IEC 62351.

[0085] 2) Based on the fault probability propagation model, the variational inference algorithm is used to perform probability inference on each short-term fault event to obtain the posterior probability of the cascading failure corresponding to the short-term fault event.

[0086] The conditional probability table in the fault probability propagation model is used to quantify the dependencies between nodes. For example, a secondary circuit break may cause the busbar differential protection to malfunction. The following joint probability distribution calculation formula can be used to quantify the dependencies between the two:

[0087]

[0088] in, Represents a set of fault events, Indicates the i-th fault event The parent node of .

[0089] Specifically, based on the conditional probability table of the fault probability propagation model, a variational inference algorithm is used to infer the probability of each short-term impending fault event, obtaining the corresponding cascading failure posterior probability. This embodiment uses a randomized variational inference algorithm to minimize the KL divergence using a mini-batch (size 1000), reducing the computational complexity of high-dimensional integrals from exponential to polynomial, improving real-time analysis efficiency.

[0090] Furthermore, if a secondary circuit disconnection is a short-term impending fault event, then a busbar differential protection misoperation is a cascading fault event. When a secondary circuit disconnection occurs, the risk score for busbar differential protection misoperation is calculated by multiplying the posterior probability of busbar differential protection misoperation P(PM|SD) by the severity of the consequence.

[0091] S4. Deducing and analyzing the dynamic topology map and the posterior probability of cascading failures to obtain several safety measures strategies, and performing multi-objective optimization sorting on each safety measure strategy to obtain a safety measure priority sequence;

[0092] Specifically, step S4 includes:

[0093] 1) Analyze the dynamic topology map and the posterior probability of cascading failures according to different risk scenarios to obtain the safety measures for each risk scenario;

[0094] The dynamic topology map ensures that the safety strategy matches the actual operating scenario, and the posterior probability of cascading failures provides data support for the differentiation of safety strategies. This embodiment deduces and analyzes the dynamic topology map and the posterior probability of cascading failures according to high-risk scenarios and low-risk scenarios, respectively, to obtain safety strategies for the corresponding risk scenarios. For high-risk scenarios where the probability of bus failure is higher than the threshold, the safety strategy is determined to forcibly enable the bus differential protection acceleration logic to shorten the fault removal time and avoid the expansion of the fault; for low-risk scenarios where the line is slightly overloaded, the safety strategy is determined to allow the backup automatic transfer device to delay the action, balancing safety and economy.

[0095] 2) Use the TOPSIS algorithm to perform multi-objective optimization on each security measure strategy and determine the priority of the corresponding security measure strategy;

[0096] Using the TOPSIS algorithm, we conduct multi-objective optimization of each safety strategy based on multi-dimensional evaluation weights to determine the priority of the corresponding safety strategy. The multi-dimensional evaluation weights include safety weight, economic weight, and recovery efficiency weight, each weight determined by the analytic hierarchy process.

[0097] 3) Sort each security measure strategy based on priority to obtain a security measure priority sequence.

[0098] Each security measure strategy is sorted in descending order based on priority to obtain a security measure priority sequence, which is the direct basis for verification execution.

[0099] It should be noted that during the verification execution phase, the present invention provides fault tolerance to prevent abnormal data from causing incorrect decisions. Specifically, three independent engines are run on physically isolated servers and synchronized using the Raft algorithm. When a result conflict occurs, the Paxos consensus algorithm is triggered to ensure decision consistency. At the same time, a conservative mode is set to lock non-critical protection, enable local cache data, and keep differential protection online when there are several consecutive communication packet losses or signal conflicts. In addition, the engine survival status is detected every second and the timeout threshold triggers an alarm, and the verification results and conflict events are recorded in the blockchain for evidence storage. In this way, a fault-tolerant mechanism for the online verification engine is constructed to ensure the reliability and security of the smart substation safety verification in abnormal scenarios.

[0100] S5. Build a digital twin of the target smart substation based on the digital mirror data of the power grid equipment, and simulate and verify the security measure priority sequence based on the digital twin to obtain a verification report;

[0101] Specifically, step S5 includes:

[0102] 1) Use digital twin technology to model and map the digital mirror data of power grid equipment in real time to obtain the digital twin of the target smart substation;

[0103] The digital mirror data for power grid equipment includes grid topology, equipment parameters, and a high-precision model library. Specifically, this embodiment uses digital twin technology to synchronize the target smart substation's grid topology and equipment parameters to the simulation environment in real time. It also integrates a high-precision model library, including the EMTP-RV electromagnetic transient model and the RT-LAB real-time simulation engine. Through data modeling and real-time mapping, dynamic synchronization between the virtual model and the physical system is achieved, thus forming a digital twin of the target smart substation.

[0104] 2) Simulate the security priority sequence based on the digital twin to obtain dynamic characteristic simulation data;

[0105] Leveraging the digital twin's real-time simulation engine, the execution of each safety strategy is simulated according to its priority sequence. During the simulation, dynamic characteristic simulation data, including electrical quantity changes and equipment status responses, is collected in real time, providing data support for subsequent safety strategy effectiveness analysis and risk assessment.

[0106] 3) Perform compliance verification on dynamic characteristic simulation data based on preset safety criteria and obtain a verification report.

[0107] The verification report includes the results of the determination of whether the operating status of the target smart substation meets the preset safety criteria after the execution of each safety strategy. Specifically, in this embodiment, the preset safety criteria is the N-1 safety criteria. Based on the dynamic characteristics simulation data, the compliance of the N-1 safety criteria is verified against the N-1 safety criteria, and a quantitative verification report is generated to prove that the N-1 safety criteria are met.

[0108] Based on the verification report obtained in step S5, the present invention provides human-machine collaborative decision-making to ensure the accuracy and interpretability of the safety measures strategy. Specifically, by designing a visual interactive interface, topology highlighting technology is used to mark high-risk and medium-risk equipment with different colors, and the posterior probability distribution of cascading failures is presented in real time with the help of an alarm panel. At the same time, an expert rule library is integrated. When the confidence level of the safety measures strategy is lower than the preset threshold, typical solutions are automatically pushed, such as citing IEC 62351 standard rules, for operation and maintenance personnel to choose to accept, modify or reject. In addition, manual intervention operations are recorded in the blockchain for evidence storage to ensure that the operations are traceable, thereby achieving deep human-machine collaboration and improving the scientificity and reliability of safety measures decision-making.

[0109] Further, if Figure 2 As shown, it is another flow chart of a method for verifying the installation measures of a smart substation according to an embodiment of the present invention. Figure 2 , after step S5, further comprising the steps of:

[0110] S6. Perform dual-loop optimization on the fault probability propagation model and the digital twin based on the verification report to obtain dual-loop feedback results.

[0111] The dual-loop optimization includes an inner loop to update the fault probability propagation model parameters and an outer loop to correct the initial state of the digital twin.

[0112] Specifically, step S6 includes:

[0113] 1) Based on the verification report, the parameters of the fault probability propagation model are optimized in the inner loop to obtain the parameters of the dynamic fault probability propagation model;

[0114] By utilizing the probabilistic reasoning ability of Bayesian networks, the inner-loop optimization of the conditional probability parameters of the fault probability propagation model is performed based on the verification report to obtain the parameters of the dynamic fault probability propagation model.

[0115] 2) Based on the verification report, the initial state of the digital twin is corrected in the outer loop to obtain the corrected initial state of the digital twin;

[0116] The digital twin, a virtual representation of the target smart substation, has a direct impact on the accuracy of its initial state. The outer loop uses the calibration report to infer and correct the initial state of the digital twin, resulting in a corrected initial state. This initial state includes device parameters and the operating environment.

[0117] 3) Based on the dynamic fault probability propagation model parameters and the initial state of the modified digital twin, a dual-loop feedback result is constructed.

[0118] S7. Optimize each security measure strategy based on the dual-loop feedback results to obtain the corresponding dynamic security measure strategy.

[0119] Specifically, based on the dual-loop feedback results, each security strategy is iteratively optimized through reinforcement learning, and the verification accuracy and verification response time are used as reward functions to obtain the corresponding dynamic security strategy.

[0120] The following describes the process of updating the security strategy through reinforcement learning in this embodiment:

[0121] 1) Define the state space;

[0122] A 12-dimensional state space is used to represent the operating state of the target smart substation. The 12-dimensional state space includes but is not limited to grid topology, protection status, and load level.

[0123] 2) Define the reward function;

[0124] The weight coefficient is used to balance the verification accuracy and verification response time, and the verification response time is normalized to the interval [0,1]. Specifically, the reward function R is represented by the following formula:

[0125] R=0.7*calibration accuracy+0.3*(1-calibration response time / 1000)

[0126] 3) Adopting algorithm update strategies;

[0127] Use the PPO (Proximal Policy Optimization) algorithm to update the strategy every 1000 steps, and the exploration rate It decays linearly from 0.3 to 0.1, balancing the exploration of new strategies and the use of existing strategies.

[0128] 4) Strategy hot update and online deployment.

[0129] By setting up a hot update interface for the strategy, new strategies can be deployed online without interrupting the target smart substation, thereby continuously optimizing the safety strategy and improving the efficiency and accuracy of safety verification.

[0130] Furthermore, to provide an unalterable audit record of the target smart substation's operations, this invention uses blockchain technology to record data from steps S1 to S7 throughout the entire process, meeting audit requirements and supporting post-event accountability. Through blockchain evidence storage, log analysis, and compliance reporting, full-process auditing and compliance management are achieved.

[0131] The embodiment of the present invention provides a method for verifying safety measures for smart substations. Based on edge federation aggregation and fault probability propagation models, it effectively integrates multi-source heterogeneous data and quantifies the risk of cascading failures, breaking through the bottleneck of traditional methods in dealing with uncertainty problems. Relying on digital mirroring of power grid equipment and optimized digital twins, it compresses simulation delays from minutes to milliseconds, enabling real-time verification of safety measures strategies. It also introduces a reinforcement learning mechanism to continuously optimize safety measures strategies and accurately adapt to new power system operating modes. The various links of the present invention work synergistically to form a complete closed loop from risk analysis, rapid verification to strategy optimization, providing efficient and intelligent technical guarantees for the safe and stable operation of smart substations.

[0132] Based on the above-mentioned smart substation security verification method, Figure 3 As shown, an embodiment of the present invention provides a smart substation safety verification system, including:

[0133] Multi-source data collection and fusion module 1 is used to collect multi-source real-time monitoring data of the target smart substation and perform edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data;

[0134] Dynamic topology perception and fault prediction module 2 is used to construct a dynamic topology map of the target smart substation based on the collaborative feature data, and perform short-term fault prediction on the target smart substation to obtain a short-term fault prediction result, wherein the short-term fault prediction result includes a number of short-term fault events and the prediction probability corresponding to each short-term fault event;

[0135] Fault Propagation Analysis Module 3 is used to construct a fault probability propagation model for the target smart substation based on the prior knowledge system and collaborative feature data, and to perform probabilistic reasoning on each short-term fault event according to the fault probability propagation model to obtain the posterior probability of cascading failures corresponding to the short-term fault event;

[0136] The emergency measure generation module 4 is used to deduce and analyze the dynamic topology map and the posterior probability of cascading failures to obtain a number of emergency measures, and perform multi-objective optimization sorting on each emergency measure strategy to obtain an emergency measure priority sequence;

[0137] The simulation verification module 5 is used to construct a digital twin of the target smart substation based on the digital mirror data of the power grid equipment, and simulate and verify the safety measure priority sequence according to the digital twin to obtain a verification report, wherein the verification report includes the judgment result of whether the operating status of the target smart substation meets the preset safety criteria after the execution of each safety measure strategy.

[0138] It should be noted that each module in the above-mentioned intelligent substation installation verification system can be fully or partially implemented by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules. For the specific definition of an intelligent substation installation verification system, please refer to the definition of an intelligent substation installation verification method above. The two have the same functions and effects and will not be repeated here.

[0139] In summary, the embodiment of the present invention provides a method and system for verifying safety measures for smart substations. Based on the edge federation aggregation and fault probability propagation model, it effectively integrates multi-source heterogeneous data and quantifies the risk of cascading failures, breaking through the bottleneck of traditional methods in dealing with uncertainty problems; relying on the digital mirror of power grid equipment and the optimized digital twin, it compresses the simulation delay from minutes to milliseconds, realizing real-time verification of safety measures strategies; introducing a reinforcement learning mechanism to continuously optimize safety measures strategies and accurately adapt to new power system operation modes. The various links of the present invention work together to form a complete closed loop from risk analysis, rapid verification to strategy optimization, providing efficient and intelligent technical guarantees for the safe and stable operation of smart substations.

[0140] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0141] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention. These improvements and substitutions should also be regarded as the scope of protection of the present invention.

Claims

1. A method for verifying the installation measures of a smart substation, characterized in that: include: Collect multi-source real-time monitoring data of the target smart substation, and perform edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data; Constructing a dynamic topology map of the target smart substation based on the collaborative feature data, and performing short-term fault prediction on the target smart substation to obtain a short-term fault prediction result, wherein the short-term fault prediction result includes a number of short-term fault events and a prediction probability corresponding to each of the short-term fault events; Constructing a fault probability propagation model of the target smart substation based on the prior knowledge system and the collaborative feature data, and performing probabilistic reasoning on each of the short-term impending fault events according to the fault probability propagation model to obtain a posterior probability of a cascading failure corresponding to the short-term impending fault event; Deducing and analyzing the dynamic topology map and the posterior probability of cascading failures to obtain a number of safety measures, and performing multi-objective optimization sorting on each of the safety measures to obtain a safety measure priority sequence; A digital twin of the target smart substation is constructed based on the digital mirror data of the power grid equipment, and the safety measure priority sequence is simulated and verified according to the digital twin to obtain a verification report, wherein the verification report includes a determination result of whether the operating status of the target smart substation meets the preset safety criteria after the execution of each safety measure strategy; Performing a dual-loop optimization on the fault probability propagation model and the digital twin based on the verification report to obtain a dual-loop feedback result, wherein the dual-loop optimization includes an inner loop updating the fault probability propagation model parameters and an outer loop correcting the initial state of the digital twin; Optimizing each of the security measures strategies based on the dual-loop feedback results to obtain a corresponding dynamic security measure strategy; The dual-loop optimization of the fault probability propagation model and the digital twin based on the verification report to obtain a dual-loop feedback result includes: performing inner-loop optimization on the parameters of the fault probability propagation model based on the verification report to obtain dynamic fault probability propagation model parameters; Performing outer-loop correction on the initial state of the digital twin based on the verification report to obtain a corrected initial state of the digital twin; Based on the dynamic fault probability propagation model parameters and the modified digital twin initial state, a dual-loop feedback result is constructed.

2. The smart substation installation verification method according to claim 1, characterized in that: The collecting of multi-source real-time monitoring data of the target smart substation and performing edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data include: Collect multi-source real-time monitoring data of the target smart substation through several edge nodes deployed in the target smart substation, wherein the multi-source real-time monitoring data includes SCADA telemetry data, PMU synchrophasor data, protection device status signals, and equipment sensor data; Preprocessing the multi-source real-time monitoring data based on each edge node to obtain local feature data corresponding to the edge node, wherein the preprocessing includes outlier removal and timestamp alignment; Each of the local feature data is aggregated based on a federated learning framework to obtain collaborative feature data.

3. The smart substation installation verification method according to claim 1, characterized in that: The constructing a dynamic topology map of the target smart substation based on the collaborative feature data, and performing short-term fault prediction on the target smart substation to obtain a short-term fault prediction result, includes: Constructing a topology map of the target smart substation based on a graph neural network, where the nodes of the topology map represent devices, and the edge weights of the topology map represent the connection reliability between devices; Updating the topology map based on the collaborative feature data to obtain a dynamic topology map; The time series status data of the target smart substation is obtained, and short-term fault prediction is performed on the time series status data based on a long short-term memory network to obtain a short-term fault prediction result.

4. The smart substation installation verification method according to claim 1, characterized in that: The method of constructing a fault probability propagation model of the target smart substation based on the prior knowledge system and the collaborative feature data, and performing probabilistic reasoning on each of the short-term fault events according to the fault probability propagation model to obtain a posterior probability of a cascading failure corresponding to the short-term fault event, includes: Based on the prior knowledge system and the collaborative feature data, a fault probability propagation model of the target smart substation is established using a Bayesian network as a framework, wherein the prior knowledge system includes a historical fault tree analysis and an expert rule base, the nodes of the fault probability propagation model represent fault events, and the directed edges of the fault probability propagation model represent causal relationships between fault events; Based on the fault probability propagation model, a variational inference algorithm is used to perform probability inference on each of the short-term impending fault events to obtain a posterior probability of a cascading failure corresponding to the short-term impending fault event.

5. The smart substation installation verification method according to claim 1, characterized in that: The dynamic topology map and the posterior probability of cascading failures are deduced and analyzed to obtain a number of safety measures, and each of the safety measures is sorted by multi-objective optimization to obtain a safety measure priority sequence, including: Deducing and analyzing the dynamic topology map and the posterior probability of cascading failures according to different risk scenarios to obtain a safety strategy for each risk scenario; Using the TOPSIS algorithm to perform multi-objective optimization on each of the security measures strategies to determine the priority of the corresponding security measures strategies; Each of the security measures strategies is sorted based on the priority to obtain a security measure priority sequence.

6. The smart substation installation verification method according to claim 1, characterized in that: The digital twin of the target smart substation is constructed based on the digital mirror data of the power grid equipment, and the security measure priority sequence is simulated and verified according to the digital twin to obtain a verification report, including: Using digital twin technology to model and map the digital mirror data of power grid equipment in real time to obtain a digital twin of the target smart substation; Simulating and executing the security measure priority sequence based on the digital twin to obtain dynamic characteristic simulation data; The dynamic characteristic simulation data is verified for compliance based on preset safety criteria to obtain a verification report.

7. The smart substation installation verification method according to claim 1, characterized in that: The step of optimizing each of the security measures based on the dual-loop feedback results to obtain a corresponding dynamic security measure strategy includes: Based on the dual-loop feedback results, each of the security measures strategies is subjected to reinforcement learning iterative optimization, and the verification accuracy and verification response time are used as reward functions to obtain the corresponding dynamic security measures strategy.

8. An intelligent substation security verification system, characterized in that: The method for verifying the installation measures of a smart substation according to any one of claims 1 to 7 is applied, wherein the system for verifying the installation measures of a smart substation comprises: A multi-source data acquisition and fusion module is used to collect multi-source real-time monitoring data of the target smart substation and perform edge federation aggregation on the multi-source real-time monitoring data to obtain collaborative feature data; a dynamic topology perception and fault prediction module, configured to construct a dynamic topology map of the target smart substation based on the collaborative feature data, and perform short-term fault prediction on the target smart substation to obtain a short-term fault prediction result, wherein the short-term fault prediction result includes a number of short-term fault events and a predicted probability corresponding to each of the short-term fault events; a fault probability propagation analysis module, configured to construct a fault probability propagation model for the target smart substation based on a priori knowledge system and the collaborative feature data, and to perform probabilistic reasoning on each of the short-term impending fault events according to the fault probability propagation model to obtain a posterior probability of a cascading failure corresponding to the short-term impending fault event; a safety measure generation module, configured to deduce and analyze the dynamic topology map and the posterior probability of cascading failures to obtain a plurality of safety measure strategies, and perform multi-objective optimization sorting on each of the safety measure strategies to obtain a safety measure priority sequence; A simulation verification module is used to construct a digital twin of the target smart substation based on the digital mirror data of the power grid equipment, and simulate and verify the safety measure priority sequence according to the digital twin to obtain a verification report, wherein the verification report includes a judgment result on whether the operating status of the target smart substation meets the preset safety criteria after the execution of each safety measure strategy.

Images