A power grid resilience improvement collaborative control method for extreme weather scenarios

Abstract

This invention relates to the field of collaborative control technology for enhancing power grid resilience, and particularly to a collaborative control method for enhancing power grid resilience in extreme weather scenarios. The method includes: collecting data to form dynamic panoramic data of the power grid; injecting the panoramic data into a digital twin model to generate quantitative early warning information including fault probability, impact range, and time window; initiating a parallel computing process based on the early warning information to obtain a set of candidate contingency plans; generating a detailed control command sequence using a distributed collaborative decision-making model; simultaneously simulating the execution of the command sequence in the digital twin model, and triggering a correction mechanism to continuously correct the commands by comparing the simulation results with the actual state in real time; recording scheduling intervention behaviors and control effects, structuring them as cases and storing them in a knowledge base, and recommending historical contingency plans based on similarity matching. This invention improves the efficiency of power grid resilience response through data fusion preprocessing, parallel distributed computing, real-time feedback correction, and knowledge reuse collaboration.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology for improving power grid resilience, and in particular to a collaborative control method for improving power grid resilience in extreme weather scenarios. Background Technology

[0002] Existing grid resilience enhancement collaborative control technologies integrate diverse resources such as distributed energy and energy storage systems to construct an adaptive control framework, enabling the grid to quickly adjust its operating status when faced with extreme weather or sudden faults. Based on advanced sensing and communication technologies, this framework achieves information sharing and action coordination among multiple entities, optimizes power flow distribution and voltage stability, thereby improving the overall robustness and recovery capability of the grid, while promoting the efficient consumption of renewable energy.

[0003] Existing collaborative control technologies for enhancing grid resilience suffer from the following technical challenges: In extreme weather scenarios, the power grid needs to rapidly integrate heterogeneous data from multiple sources, such as meteorological satellites, line sensors, and load forecasting units, for situational awareness. However, due to differences in data formats, sampling frequencies, and communication protocols, achieving low-latency, high-reliability data fusion is inherently challenging. The latency of multi-source data fusion directly leads to a delay in the grid operating status information acquired by the collaborative decision-making model. Furthermore, this collaborative decision-making model requires online coordination and optimization of various strategies, such as load shedding, energy storage charging and discharging, generator output, and network topology reconfiguration, forming a high-dimensional, nonlinear mixed-integer programming problem, the computational complexity of which increases dramatically with the scale of the power grid. For example, when a typhoon causes successive trips of main lines, the dispatch center cannot grasp the power flow trend of the entire network in real time due to slow data fusion. At the same time, the decision-making model struggles to select the optimal control sequence that balances safety constraints and economy from a massive number of feasible solutions within a limited time. Ultimately, the generated key control command set lags significantly behind the dynamic evolution of the power grid fault, weakening the collaborative control system's ability to respond quickly to extreme events. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a collaborative control method for enhancing power grid resilience in extreme weather scenarios. This method solves the technical problem of delayed generation of control commands for enhancing power grid resilience caused by the difficulty in real-time fusion of multi-source data and the computational complexity of collaborative decision-making models under extreme weather conditions.

[0005] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows:

[0006] This invention provides a collaborative control method for enhancing power grid resilience in extreme weather scenarios, comprising:

[0007] Step 1: Collect real-time electrical quantity data and weather forecast data of the power grid, attach high-precision time labels to all data, and overlay and fuse the electrical quantity data and weather forecast data in the spatial dimension to obtain dynamic panoramic data of the power grid.

[0008] Step 2: Inject the dynamic panoramic data of the power grid into the digital twin model of the power grid. Simulate the fault evolution process of power grid components under extreme weather events through the digital twin model, dynamically evaluate the power grid stability indicators, and generate quantitative early warning information, which includes fault probability, impact range and time window.

[0009] Step 3: Based on the quantified early warning information, start multiple parallel computing processes. Each process corresponds to an extreme weather development scenario to build a multi-objective optimization model, solve for a set of preliminary control schemes, perform cross-scenario cross-validation on the set of preliminary control schemes, and select a set of candidate contingency plans.

[0010] Step 4: Select a basic scheme from the candidate scheme set, construct a distributed collaborative decision-making model, solve it using a distributed optimization algorithm, and generate a detailed control instruction sequence;

[0011] Step 5: Simultaneously send the detailed control command sequence to the grid execution device and pre-install it into the grid digital twin model. Simulate the execution of the control command sequence through the digital twin model, compare the key stability indicators in the simulation results with the actual operating state of the grid, and trigger the correction mechanism to perform rolling correction on subsequent unexecuted commands when the deviation exceeds the preset threshold.

[0012] Step 6: During the execution of control commands, receive intervention from dispatchers through the human-machine interface, record the intervention behavior and control effect, and store the data as a case in the contingency plan knowledge base. When encountering similar extreme weather scenarios again, recommend historical contingency plans from the contingency plan knowledge base.

[0013] Furthermore, the collaborative control method for enhancing power grid resilience in extreme weather scenarios described in this invention includes the following steps: collecting real-time electrical quantity data and meteorological forecast data of the power grid, which includes:

[0014] Voltage phase, frequency, and power flow data of power grid nodes are collected using a synchronous phasor measurement device.

[0015] The circuit breaker status and transformer tap position information are obtained through a data acquisition and monitoring system.

[0016] Receive gridded weather forecast data for the next 6 to 72 hours through a meteorological data interface;

[0017] High-precision timestamps from the BeiDou / GPS system are added to the collected voltage phase, frequency, power flow data, circuit breaker status, transformer tap location information, and gridded weather forecast data.

[0018] Furthermore, the collaborative control method for enhancing power grid resilience in extreme weather scenarios described in this invention, wherein the spatial overlay and fusion of electrical quantity data and meteorological forecast data includes:

[0019] The spatial mapping function of the power geographic information system is invoked to map time-stamped electrical quantity data and meteorological elements to the power grid topology map based on latitude and longitude coordinates;

[0020] The electrical quantity data mapped to the power grid topology is aligned with the meteorological elements in a spatiotemporal manner, and the resulting data is then fused to output a dynamic panoramic data map of the power grid.

[0021] Furthermore, the collaborative control method for enhancing grid resilience in extreme weather scenarios described in this invention includes, in simulating the fault evolution process of grid components under extreme weather events, the following steps:

[0022] The Monte Carlo stochastic simulation method was used to repeatedly simulate the evolution of extreme weather in a digital twin environment;

[0023] Based on the simulated extreme weather evolution process, a component failure sequence that conforms to historical statistical patterns is randomly generated, and the system transient stability index, voltage stability margin and line overload conditions are calculated.

[0024] Based on the calculated system transient stability index, voltage stability margin, and line overload conditions, the probability of failure and the risk of cascading failures for different lines and substations in different time periods are statistically derived.

[0025] Furthermore, the collaborative control method for enhancing grid resilience in extreme weather scenarios described in this invention includes cross-scenario verification of the preliminary control scheme set, comprising:

[0026] The preferred preliminary control scheme generated in the first process is substituted into the extreme weather development scenario model corresponding to the second process for simulation verification.

[0027] The better-performing preparatory control scheme in the second process is substituted into the extreme weather development scenario model of the first process for an adaptability test.

[0028] Based on the results of simulation verification and adaptability testing, a set of candidate contingency plans that perform stably under various potential risk scenarios were selected.

[0029] Furthermore, the collaborative control method for improving power grid resilience in extreme weather scenarios described in this invention, wherein the solution is obtained using a distributed optimization algorithm, includes:

[0030] The power grid is divided into multiple autonomous regions according to the degree of topological coupling, and a local decision-making agent is assigned to each region.

[0031] Each regional decision agent solves for the optimal solution of local control variables in parallel, including generator output adjustment, load reduction schemes, and network topology change sequences;

[0032] A consistency coordination mechanism for inter-regional boundary coupling constraints is established using the alternating direction multiplier method, and the Lagrange multipliers are iteratively updated.

[0033] When the power flow deviation at the boundary of adjacent regions is less than the convergence threshold, a globally coordinated and consistent detailed control command sequence is output.

[0034] The solution process of the distributed optimization algorithm is synchronized with the real-time state update of the digital twin model.

[0035] Furthermore, the collaborative control method for enhancing grid resilience in extreme weather scenarios described in this invention includes, in part, comparing key stability indicators from simulation results with the actual operating state of the power grid, comprising:

[0036] Construct a forward-looking rolling time window in a digital twin environment to simulate the execution of a sequence of control commands for the next 15-30 minutes;

[0037] Real-time tracking of voltage trajectories at central nodes, power fluctuations at key sections, and dynamic frequency response characteristics of the system;

[0038] Establish a dynamic error propagation model between actual measurement data and twin predictions. When the prediction deviation exceeds the adaptive threshold or trend instability indicator, trigger the correction mechanism.

[0039] The correction mechanism feeds back the deviation information to the distributed collaborative decision-making model, initiating a re-optimization process of the instruction sequence based on the current actual state.

[0040] Furthermore, in the collaborative control method for enhancing grid resilience in extreme weather scenarios described in this invention, the step of recommending historical contingency plans from a contingency plan knowledge base includes:

[0041] Extract meteorological intensity characteristics, power grid operation topology characteristics, and load distribution pattern characteristics of the current extreme weather scenario to construct a case feature vector space;

[0042] The dynamic time warping algorithm is used to calculate the multi-dimensional similarity between the feature vector of the current extreme weather scenario and the feature vector of historical cases in the contingency plan knowledge base;

[0043] When the calculated similarity exceeds the matching threshold, the top K historical cases with the best control effect are retrieved from the contingency plan knowledge base;

[0044] The contingency plans and strategies from retrieved historical cases are adaptively modified to reflect the current real-time situation, generating a weighted and fused recommended solution.

[0045] The weighted fusion recommendation scheme is used as the initial population input to the multi-objective optimization algorithm in the parallel computing process.

[0046] Furthermore, the collaborative control method for enhancing power grid resilience in extreme weather scenarios described in this invention includes generating quantitative early warning information comprising fault probability, impact range, and time window, comprising:

[0047] Based on the inference results of the digital twin model, a spatiotemporally correlated chain propagation map of power grid vulnerability is constructed.

[0048] Identify key weak links and their fault propagation paths from the chain propagation map of power grid vulnerability, and mark the failure probability time series of each link;

[0049] Based on the identified key weak links, fault propagation paths, and marked failure probability time series, an early warning information matrix is ​​generated. The dimensions of the early warning information matrix include component identification, risk level, time window, and impact weight.

[0050] The early warning information matrix serves as an initialization parameter for the parallel computing process, used to configure the optimization model constraints for each process.

[0051] Furthermore, in the collaborative control method for enhancing power grid resilience in extreme weather scenarios described in this invention, the early warning information matrix serves as an initialization parameter for the parallel computing process, used to configure the optimization model constraints for each process, including:

[0052] The early warning information matrix is ​​input into the parallel computing process as the initialization parameter of the optimization model;

[0053] Based on the component identifiers in the early warning information matrix, set the operational constraints for the corresponding components in the optimization model;

[0054] Based on the risk level and time window in the early warning information matrix, set the safety constraints in the optimization model;

[0055] Based on the influence weights in the early warning information matrix, set the objective function weight parameters of the optimization model.

[0056] Beneficial effects of this invention;

[0057] This invention effectively solves the problems of difficult real-time fusion of multi-source data and control command generation delays caused by the computational complexity of collaborative decision-making models under extreme weather conditions by integrating data fusion, digital twin simulation, parallel distributed computing, and real-time feedback correction technologies. In the data acquisition phase, a synchronous phasor measurement device is used to collect electrical quantity data of power grid nodes, and equipment status information is obtained through a data acquisition and monitoring system. Simultaneously, gridded weather forecast data is received via a meteorological data interface. High-precision timestamps are added to all data to achieve time synchronization. Then, a power geographic information system is invoked for spatial mapping, aligning the electrical quantity data and meteorological elements to the power grid topology map based on latitude and longitude coordinates, forming a spatiotemporally consistent dynamic panoramic data of the power grid, eliminating the fusion delay caused by data heterogeneity. The dynamic panoramic data of the power grid drives the digital twin model to perform Monte Carlo stochastic simulation, deduce the chain propagation process of component failures under extreme weather conditions, and generate quantitative early warning information including failure probability, impact range, and time window, providing accurate input for decision-making. Based on the quantitative early warning information, multiple parallel computing processes are initiated, each targeting a specific type of extreme weather event. A multi-objective optimization model is constructed to solve for a set of preliminary control schemes based on weather development scenarios. Candidate schemes are then screened through cross-scenario cross-validation, reducing the decision search space. A distributed collaborative decision-making model is adopted, dividing the power grid into autonomous regions based on topological coupling. Each region solves its local control variables in parallel using the alternating direction multiplier method, achieving global optimization only through boundary power flow coordination, thus reducing computational complexity. During the command execution phase, a digital twin model performs look-ahead simulations, comparing the simulation results of the control command sequence with actual power grid measurement data in real time. When the deviation exceeds a threshold, a correction mechanism is triggered to continuously correct unexecuted commands. Simultaneously, dispatcher intervention behaviors and control effects are recorded through a human-machine interface, structured into cases and stored in the scheme knowledge base. When similar extreme weather scenarios recur, historical schemes are recommended as the initial population for parallel computation through feature vector similarity matching, accelerating optimization convergence. This technological chain, through deep collaboration of data preprocessing, parallel computing, real-time correction, and knowledge reuse, achieves full-process acceleration from situational awareness to decision execution, significantly improving the power grid's resilience response capability under extreme weather conditions. Attached Figure Description

[0058] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the drawings without creative effort.

[0059] Figure 1 This is a flowchart of a collaborative control method for improving power grid resilience in extreme weather scenarios, according to the present invention. Detailed Implementation

[0060] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0061] Please see Figure 1 This invention provides a collaborative control method for enhancing power grid resilience in extreme weather scenarios, comprising:

[0062] Step 1: Collect real-time electrical quantity data and weather forecast data of the power grid, attach high-precision time labels to all data, and overlay and fuse the electrical quantity data and weather forecast data in the spatial dimension to obtain dynamic panoramic data of the power grid.

[0063] Step 2: Inject the dynamic panoramic data of the power grid into the digital twin model of the power grid. Simulate the fault evolution process of power grid components under extreme weather events through the digital twin model, dynamically evaluate the power grid stability indicators, and generate quantitative early warning information, which includes fault probability, impact range and time window.

[0064] Step 3: Based on the quantified early warning information, start multiple parallel computing processes. Each process corresponds to an extreme weather development scenario to build a multi-objective optimization model, solve for a set of preliminary control schemes, perform cross-scenario cross-validation on the set of preliminary control schemes, and select a set of candidate contingency plans.

[0065] Step 4: Select a basic scheme from the candidate scheme set, construct a distributed collaborative decision-making model, solve it using a distributed optimization algorithm, and generate a detailed control instruction sequence;

[0066] Step 5: Simultaneously send the detailed control command sequence to the grid execution device and pre-install it into the grid digital twin model. Simulate the execution of the control command sequence through the digital twin model, compare the key stability indicators in the simulation results with the actual operating state of the grid, and trigger the correction mechanism to perform rolling correction on subsequent unexecuted commands when the deviation exceeds the preset threshold.

[0067] Step 6: During the execution of control commands, receive intervention from dispatchers through the human-machine interface, record the intervention behavior and control effect, and store the data as a case in the contingency plan knowledge base. When encountering similar extreme weather scenarios again, recommend historical contingency plans from the contingency plan knowledge base.

[0068] This invention provides a collaborative control method for enhancing power grid resilience in extreme weather scenarios, specifically including the following detailed implementation steps. First, real-time electrical quantity data and weather forecast data of the power grid are collected. High-precision time tags are added to all data, and the electrical quantity data and weather forecast data are spatially overlaid and fused to obtain dynamic panoramic data of the power grid. Electrical quantity data is collected from the voltage phase, frequency, and power flow data of power grid nodes through a synchronous phasor measurement device, and circuit breaker status and transformer tap position information are obtained through a data acquisition and monitoring system. Weather forecast data is received through a meteorological data interface, receiving gridded weather forecast data for a future period. All collected data are appended with high-precision timestamps from the BeiDou or GPS systems to achieve time synchronization. Subsequently, the spatial mapping function of the power geographic information system is invoked to map the time-tagged electrical quantity data and meteorological elements onto the power grid topology map based on latitude and longitude coordinates. The mapped data is then aligned and fused using spatiotemporal references to output a dynamic panoramic data map of the power grid.

[0069] Next, dynamic panoramic data of the power grid is injected into the power grid digital twin model. This model simulates the fault evolution process of power grid components under extreme weather events, dynamically assesses power grid stability indicators, and generates quantitative early warning information. This quantitative early warning information includes fault probability, impact range, and time window. In the power grid digital twin model, the Monte Carlo stochastic simulation method is used to repeatedly simulate the evolution process of extreme weather. Based on the simulated extreme weather evolution process, component fault sequences conforming to historical statistical patterns are randomly generated. System transient stability indicators, voltage stability margin, and line overload conditions are calculated. Based on the calculated system transient stability indicators, voltage stability margin, and line overload conditions, the fault probability and cascading fault risk of different lines and substations in future time periods are statistically derived.

[0070] Based on quantitative early warning information, multiple parallel computing processes are initiated. Each process corresponds to an extreme weather development scenario, constructing a multi-objective optimization model to obtain a set of preliminary control schemes. These preliminary control schemes are then cross-validated across scenarios to select a set of candidate contingency plans. Each parallel computing process, for an extreme weather development scenario defined in the quantitative early warning information, constructs a robust optimization model with the objectives of minimizing load shedding, minimizing control costs, and maximizing system stability margin. A multi-objective optimization algorithm, such as the NSGA-II algorithm, is used to solve this model, obtaining a set of Pareto optimal solutions as preliminary control schemes. Subsequently, cross-scenario cross-validation is performed. The preferred preliminary control schemes generated in the first process are substituted into the extreme weather development scenario model corresponding to the second process for simulation verification. Simultaneously, the better-performing preliminary control schemes from the second process are substituted into the extreme weather development scenario model of the first process for adaptability testing. Based on the simulation verification and adaptability testing results, a set of candidate contingency plans that demonstrate stability under various potential risk scenarios is selected.

[0071] A basic scheme is selected from the candidate scheme set, a distributed collaborative decision-making model is constructed, and a distributed optimization algorithm is used to solve the problem, generating a detailed control command sequence. Dispatchers select a basic scheme from the candidate scheme set based on real-time operational status, and then divide the power grid into multiple autonomous regions according to topological coupling, assigning a local decision agent to each region. Each region's decision agent solves for the optimal solutions to local control variables in parallel, including generator output adjustments, load reduction schemes, and network topology change sequences. A consistency coordination mechanism for inter-regional boundary coupling constraints is established using the alternating direction multiplier method, iteratively updating the Lagrange multipliers. When the power flow deviation between adjacent regions is less than the convergence threshold, a globally coordinated detailed control command sequence is output. The solution process of the distributed optimization algorithm is synchronized with the real-time status update of the power grid digital twin model.

[0072] Detailed control command sequences are simultaneously sent to the grid execution devices and pre-loaded into the grid digital twin model. The grid digital twin model simulates the execution of the control command sequences, and the key stability indicators in the simulation results are compared with the actual operating state of the grid. When the deviation exceeds a preset threshold, a correction mechanism is triggered to continuously correct subsequent unexecuted commands. The control command sequences are sent to field execution units such as power plants and energy storage stations, and simultaneously fed into the grid digital twin model for forward simulation, simulating the dynamic process of the system over a future period and predicting key stability indicators such as central node voltage and power fluctuations at important sections. The system continuously compares the predicted values ​​of the grid digital twin model with the real-time data transmitted from the actual measurement system, establishing a dynamic error propagation model between the actual measurement data and the twin prediction values. When the prediction deviation exceeds an adaptive threshold or a trend instability indicator, a correction mechanism is triggered, feeding the deviation information back to the distributed collaborative decision-making model and initiating a command sequence re-optimization process based on the current actual state.

[0073] During the execution of control commands, the system receives interventions from dispatchers through a human-machine interface, records intervention behaviors and control effects, and structures these into cases stored in the contingency plan knowledge base. When similar extreme weather scenarios are encountered again, historical contingency plans are recommended from the contingency plan knowledge base. Dispatchers monitor the control effects and system status in real time through a 3D visualization human-machine interface, and can manually fine-tune or confirm at key decision nodes. The system records intervention behaviors and the final control effects. Complete event data, including early warning information, decision-making process, execution results, and correction records, is structured into cases and stored in the contingency plan knowledge base. When a new extreme weather scenario occurs, the system extracts the meteorological intensity characteristics, power grid operation topology characteristics, and load distribution pattern characteristics of the current scenario to construct a case feature vector space. A dynamic time warping algorithm is used to calculate the multi-dimensional similarity between the current scenario feature vector and the historical case feature vector. When the similarity exceeds the matching threshold, the system retrieves the historical case with the best control effect from the contingency plan knowledge base, adaptively modifies the contingency plan strategy of the historical case with the current real-time situation, and generates a weighted fusion recommendation scheme, which is used as the initial population input to the multi-objective optimization algorithm for the parallel computing process.

[0074] When collecting real-time electrical quantity data and weather forecast data of the power grid, this invention specifically collects voltage phase, frequency, and power flow data of power grid nodes through a synchronous phasor measurement device. At the same time, it obtains circuit breaker status and transformer tap position information through a data acquisition and monitoring system, and receives gridded weather forecast data for a period of time in the future through a meteorological data interface. Subsequently, a high-precision timestamp from the Beidou system or GPS system is added to all collected data to achieve data time synchronization and provide a unified time reference for subsequent data fusion.

[0075] When this invention overlays and fuses electrical quantity data and meteorological forecast data in a spatial dimension, it specifically calls the spatial mapping function of the power geographic information system to map the time-labeled electrical quantity data and meteorological elements to the power grid topology map based on latitude and longitude coordinates. Then, the electrical quantity data and meteorological elements mapped to the power grid topology map are aligned with the spatiotemporal reference to eliminate spatial and temporal differences. Finally, the dynamic panoramic data map of the power grid is fused and output to form a unified power grid status view.

[0076] In simulating the fault evolution process of power grid components under extreme weather events, this invention specifically employs the Monte Carlo stochastic simulation method to repeatedly simulate the extreme weather evolution process in a digital twin environment. Based on the simulated extreme weather evolution process, it randomly generates component fault sequences that conform to historical statistical patterns and calculates system transient stability indicators, voltage stability margin, and line overload conditions. Based on the calculation results, it statistically derives the fault probability and cascading fault risk of different lines and substations in future time periods, thereby generating quantitative early warning information.

[0077] When performing cross-scenario verification of the set of preliminary control schemes, this invention specifically substitutes the preferred preliminary control schemes generated in the first process into the extreme weather development scenario model corresponding to the second process for simulation verification. At the same time, it substitutes the preliminary control schemes with better evaluation in the second process into the extreme weather development scenario model of the first process for adaptability testing. Based on the results of simulation verification and adaptability testing, a set of candidate contingency plans that perform stably under multiple potential risk scenarios is selected to improve the robustness of the contingency plans.

[0078] When using a distributed optimization algorithm, this invention specifically divides the power grid into multiple autonomous regions according to the degree of topological coupling, and assigns a local decision agent to each region. Each region's decision agent solves the optimal solution for local control variables in parallel, including generator output adjustment, load reduction scheme, and network topology change sequence. A consistency coordination mechanism for boundary coupling constraints between regions is established through the alternating direction multiplier method, and the Lagrange multipliers are iteratively updated. When the power flow deviation between adjacent regions is less than the convergence threshold, a globally coordinated and consistent detailed control command sequence is output, and it maintains synchronous interaction with the real-time state update of the digital twin model.

[0079] When comparing key stability indicators in the simulation results with the actual operating state of the power grid, this invention specifically constructs a forward-looking rolling time window in the digital twin environment to simulate the execution of control command sequences over a future period; it tracks the voltage trajectory of central nodes, power fluctuations at key sections, and dynamic frequency response characteristics of the system in real time; it establishes a dynamic error propagation model between actual measurement data and twin predictions, and triggers a correction mechanism when the prediction deviation exceeds an adaptive threshold or trend instability indicator; the correction mechanism feeds back the deviation information to the distributed collaborative decision-making model, initiating a command sequence re-optimization process based on the current actual state.

[0080] When recommending historical contingency plans from the contingency plan knowledge base, this invention specifically extracts the meteorological intensity characteristics, power grid operation topology characteristics, and load distribution pattern characteristics of the current extreme weather scenario to construct a case feature vector space. A dynamic time warping algorithm is used to calculate the multi-dimensional similarity between the feature vector of the current extreme weather scenario and the feature vectors of historical cases in the contingency plan knowledge base. When the similarity exceeds a matching threshold, the historical case with the best control effect is retrieved from the contingency plan knowledge base. The contingency plan strategy of the retrieved historical case is adaptively modified with the current real-time situation to generate a weighted fusion recommendation scheme, which is then used as the initial population input to a multi-objective optimization algorithm in the parallel computing process.

[0081] This invention, when generating quantitative early warning information including fault probability, impact range, and time window, specifically constructs a spatiotemporally correlated power grid vulnerability chain propagation map based on the deduction results of a digital twin model; identifies key weak links and their fault propagation paths from the power grid vulnerability chain propagation map, and marks the failure probability time series of each link; generates an early warning information matrix based on the identified key weak links, fault propagation paths, and marked failure probability time series. The dimensions of the early warning information matrix include component identification, risk level, time window, and impact weight; the early warning information matrix serves as the initialization parameter for the parallel computing process, used to configure the optimization model constraints of each process.

[0082] This invention effectively solves the problems of difficulty in real-time fusion of multi-source data under extreme weather conditions and instruction generation delays caused by the computational complexity of collaborative decision-making models through a series of collaborative technical means. First, in the data acquisition phase, a synchronous phasor measurement device and a data acquisition and monitoring system are used to acquire real-time electrical quantity data of the power grid. Simultaneously, gridded weather forecast data is received through a meteorological data interface, and high-precision timestamps are added to all data to achieve time synchronization of multi-source data. Then, a power geographic information system is invoked for spatial mapping, aligning the electrical quantity data and meteorological elements to the power grid topology map based on latitude and longitude coordinates, forming a spatiotemporally consistent dynamic panoramic data of the power grid, fundamentally eliminating the fusion delay caused by data heterogeneity. Next, the dynamic panoramic data of the power grid drives a digital twin model to perform Monte Carlo stochastic simulations, deduce the chain propagation process of component failures under extreme weather conditions, and generate quantitative early warnings including probabilistic risk information, providing accurate input for subsequent decision-making and avoiding decision lags caused by incomplete data in existing methods.

[0083] To address the computational complexity of collaborative decision-making models, this invention employs a parallel computing architecture. Multiple computational processes are initiated based on quantified early warning information. Each process corresponds to a specific extreme weather scenario, constructing a multi-objective optimization model and independently solving for a set of preliminary control schemes. Through a cross-scenario cross-validation mechanism, the optimal schemes from different processes are mutually substituted for verification, selecting a set of candidate schemes robust across scenarios, significantly reducing the search space required for centralized optimization. Furthermore, a distributed collaborative decision-making model is adopted, dividing the power grid into autonomous regions based on topological coupling. Each region solves its local control variables in parallel using the alternating direction multiplier method, achieving global optimization only through boundary power flow coordination. This decomposes the high-dimensional mixed-integer programming problem into multiple parallelizable subproblems, greatly reducing computational complexity.

[0084] During the command execution phase, a forward-looking simulation is performed using a digital twin model. The simulation results of the control command sequence are compared in real time with actual power grid measurement data. When the deviation exceeds a threshold, a correction mechanism is triggered to dynamically adjust unexecuted commands, forming a closed-loop control. Simultaneously, the human-machine interface records dispatcher intervention behaviors and control effects, which are structured into cases and stored in the contingency plan knowledge base. When similar extreme weather scenarios recur, the best historical contingency plan is quickly recommended through feature vector similarity matching, serving as the initial population for parallel computing and accelerating optimization convergence. The entire method achieves end-to-end acceleration from data perception to decision execution through the collaborative efforts of four layers of technology: data fusion preprocessing, parallel distributed computing, real-time feedback correction, and knowledge reuse, fundamentally solving the technical challenge of command generation delay.

Claims

1. A collaborative control method for enhancing power grid resilience in extreme weather scenarios, characterized in that, include: Step 1: Collect real-time electrical quantity data and weather forecast data of the power grid, attach high-precision time labels to all data, and overlay and fuse the electrical quantity data and weather forecast data in the spatial dimension to obtain dynamic panoramic data of the power grid. Step 2: Inject the dynamic panoramic data of the power grid into the digital twin model of the power grid. Simulate the fault evolution process of power grid components under extreme weather events through the digital twin model, dynamically evaluate the power grid stability indicators, and generate quantitative early warning information, which includes fault probability, impact range and time window. Step 3: Based on the quantified early warning information, start multiple parallel computing processes. Each process corresponds to an extreme weather development scenario to build a multi-objective optimization model, solve for a set of preliminary control schemes, perform cross-scenario cross-validation on the set of preliminary control schemes, and select a set of candidate contingency plans. Step 4: Select a basic scheme from the candidate scheme set, construct a distributed collaborative decision-making model, solve it using a distributed optimization algorithm, and generate a detailed control instruction sequence; Step 5: Simultaneously send the detailed control command sequence to the grid execution device and pre-install it into the grid digital twin model. Simulate the execution of the control command sequence through the digital twin model, compare the key stability indicators in the simulation results with the actual operating state of the grid, and trigger the correction mechanism to perform rolling correction on subsequent unexecuted commands when the deviation exceeds the preset threshold. Step 6: During the execution of control commands, receive intervention from dispatchers through the human-machine interface, record the intervention behavior and control effect, and store the data as a case in the contingency plan knowledge base. When encountering similar extreme weather scenarios again, recommend historical contingency plans from the contingency plan knowledge base.

2. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 1, characterized in that, The collection of real-time electrical quantity data and meteorological forecast data of the power grid includes: Voltage phase, frequency, and power flow data of power grid nodes are collected using a synchronous phasor measurement device. The circuit breaker status and transformer tap position information are obtained through a data acquisition and monitoring system. Receive gridded weather forecast data for the next 6 to 72 hours through a meteorological data interface; High-precision timestamps from the BeiDou / GPS system are added to the collected voltage phase, frequency, power flow data, circuit breaker status, transformer tap location information, and gridded weather forecast data.

3. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 2, characterized in that, The process of spatially overlaying and fusing electrical quantity data with meteorological forecast data includes: The spatial mapping function of the power geographic information system is invoked to map time-stamped electrical quantity data and meteorological elements to the power grid topology map based on latitude and longitude coordinates; The electrical quantity data mapped to the power grid topology is aligned with the meteorological elements in a spatiotemporal manner, and the resulting data is then fused to output a dynamic panoramic data map of the power grid.

4. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 3, characterized in that, The simulated fault evolution process of power grid components under extreme weather events includes: The Monte Carlo stochastic simulation method was used to repeatedly simulate the evolution of extreme weather in a digital twin environment; Based on the simulated extreme weather evolution process, a component failure sequence that conforms to historical statistical patterns is randomly generated, and the system transient stability index, voltage stability margin and line overload conditions are calculated. Based on the calculated system transient stability index, voltage stability margin, and line overload conditions, the probability of failure and the risk of cascading failures for different lines and substations in different time periods are statistically derived.

5. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 4, characterized in that, The cross-scenario verification of the preliminary control scheme set includes: The preferred preliminary control scheme generated in the first process is substituted into the extreme weather development scenario model corresponding to the second process for simulation verification. The better-performing preparatory control scheme in the second process is substituted into the extreme weather development scenario model of the first process for an adaptability test. Based on the results of simulation verification and adaptability testing, a set of candidate contingency plans that perform stably under various potential risk scenarios were selected.

6. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 5, characterized in that, The solution obtained using a distributed optimization algorithm includes: The power grid is divided into multiple autonomous regions according to the degree of topological coupling, and a local decision-making agent is assigned to each region. Each regional decision agent solves for the optimal solution of local control variables in parallel, including generator output adjustment, load reduction schemes, and network topology change sequences; A consistency coordination mechanism for inter-regional boundary coupling constraints is established using the alternating direction multiplier method, and the Lagrange multipliers are iteratively updated. When the power flow deviation at the boundary of adjacent regions is less than the convergence threshold, a globally coordinated and consistent detailed control command sequence is output. The solution process of the distributed optimization algorithm is synchronized with the real-time state update of the digital twin model.

7. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 6, characterized in that, The comparison of key stability indicators in the simulation results with the actual operating state of the power grid includes: Construct a forward-looking rolling time window in a digital twin environment to simulate the execution of a sequence of control commands for the next 15-30 minutes; Real-time tracking of voltage trajectories at central nodes, power fluctuations at key sections, and dynamic frequency response characteristics of the system; Establish a dynamic error propagation model between actual measurement data and twin predictions. When the prediction deviation exceeds the adaptive threshold or trend instability indicator, trigger the correction mechanism. The correction mechanism feeds back the deviation information to the distributed collaborative decision-making model, initiating a re-optimization process of the instruction sequence based on the current actual state.

8. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 7, characterized in that, The recommendation of historical contingency plans from the contingency plan knowledge base includes: Extract meteorological intensity characteristics, power grid operation topology characteristics, and load distribution pattern characteristics of the current extreme weather scenario to construct a case feature vector space; The dynamic time warping algorithm is used to calculate the multi-dimensional similarity between the feature vector of the current extreme weather scenario and the feature vector of historical cases in the contingency plan knowledge base; When the calculated similarity exceeds the matching threshold, the top K historical cases with the best control effect are retrieved from the contingency plan knowledge base; The contingency plans and strategies from retrieved historical cases are adaptively modified to reflect the current real-time situation, generating a weighted and fused recommended solution. The weighted fusion recommendation scheme is used as the initial population input to the multi-objective optimization algorithm in the parallel computing process.

9. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 8, characterized in that, The generation of quantitative early warning information includes fault probability, impact range, and time window, including: Based on the inference results of the digital twin model, a spatiotemporally correlated chain propagation map of power grid vulnerability is constructed. Identify key weak links and their fault propagation paths from the chain propagation map of power grid vulnerability, and mark the failure probability time series of each link; Based on the identified key weak links, fault propagation paths, and marked failure probability time series, an early warning information matrix is ​​generated. The dimensions of the early warning information matrix include component identification, risk level, time window, and impact weight. The early warning information matrix serves as an initialization parameter for the parallel computing process, used to configure the optimization model constraints for each process.

10. The collaborative control method for enhancing power grid resilience in extreme weather scenarios according to claim 9, characterized in that, The early warning information matrix serves as an initialization parameter for the parallel computing process, used to configure the optimization model constraints for each process, including: The early warning information matrix is ​​input into the parallel computing process as the initialization parameter of the optimization model; Based on the component identifiers in the early warning information matrix, set the operational constraints for the corresponding components in the optimization model; Based on the risk level and time window in the early warning information matrix, set the safety constraints in the optimization model; Based on the influence weights in the early warning information matrix, set the objective function weight parameters of the optimization model.

Images