Photovoltaic load regulation and control method and system based on fuzzy matching and topology dynamic optimization

Abstract

The invention discloses a photovoltaic load regulation and control method and system based on fuzzy matching and topology dynamic optimization. The method comprises the following steps: S1, presetting power grid state data acquisition equipment; s2, performing single-node discrete feature vector and dynamic power grid state tensor construction on the state data of each node of the power grid and the distributed photovoltaic system; s3, constructing a reference power grid state tensor based on the historical data of the single node of the power grid; s4, eliminating abnormal dynamic power grid state tensors in each single node of the power grid; s5, performing multi-source power grid data matching based on the power grid node data topological structure and fuzzy calculation; s6, verifying and optimizing the multi-source power grid data based on a multi-source power grid data matching result; s7, storing a multi-source power grid data version, and storing a power grid state data section; and S8, performing multi-copy management and simulation support based on all the multi-source data verified and optimized in the step S6 and the data end face stored in the step S7. According to the method, the fluctuation and uncertainty of distributed photovoltaic power generation can be regulated and controlled, and autonomous consumption in a region is realized.

Description

Technical Field

[0001] The present invention relates to the field of power grid data verification, and in particular to a photovoltaic load control method and system based on fuzzy matching and topology dynamic optimization. Background Art

[0002] The current grid dispatch verification mainly relies on traditional monitoring and analysis methods, which have several significant defects. First, most traditional grid monitoring equipment is distributed in fixed locations, with a single data source, and cannot effectively integrate multi-source data from different devices and regions. This leads to incomplete data coverage and cannot fully reflect the actual operating status of the grid. Especially with the widespread application of distributed photovoltaic systems, traditional methods have become more difficult to obtain and integrate data from these new distributed energy sources in real time, making it impossible to accurately assess their impact on the grid.

[0003] Secondly, the existing verification methods mostly use static feature matching and analysis methods. These methods perform poorly when processing dynamically changing data, and it is difficult to reflect the real-time changes in the power grid status in a timely manner, resulting in delayed scheduling decisions. For distributed photovoltaic systems, the hysteresis problem of traditional methods is particularly prominent because their power generation is affected by light conditions and changes rapidly. In addition, traditional methods have insufficient processing capabilities for data outliers and noise, and are easily affected by abnormal data, resulting in inaccurate data analysis results. In distributed photovoltaic systems, due to their dispersed equipment and complex environment, abnormal data is more likely to be generated, which further exacerbates the severity of the problem.

[0004] Furthermore, in terms of data storage and version management, existing technologies lack effective version control and historical data management mechanisms. Grid dispatch data is usually stored in a decentralized manner, making it difficult to manage and track versions in a unified manner. This makes it impossible to guarantee the integrity and traceability of data, posing a huge challenge to dispatch optimization and fault handling. For distributed photovoltaic systems, data management is faced with the problem of numerous devices and large amounts of data, which is difficult to cope with with traditional storage methods.

[0005] The above deficiencies lead to the fact that the safety and reliability of power grid dispatching cannot meet the requirements of modern power grids for efficient and accurate dispatching. Therefore, there is an urgent need for a new photovoltaic load control method that can integrate multi-source data, monitor the power grid status in real time, and dynamically optimize and version the data. In particular, it is necessary to fully consider the data integration and management needs of distributed photovoltaic systems to improve the overall efficiency and reliability of power grid dispatching. Summary of the invention

[0006] The purpose of the present invention is to overcome the above-mentioned defects of the prior art and propose a photovoltaic load control method and system based on fuzzy matching and topology dynamic optimization. The present invention overcomes the problems of data inconsistency, imperfect version control, and poor real-time performance in traditional photovoltaic load control methods. By introducing fuzzy matching and topology dynamic optimization algorithms, the accuracy of multi-source data matching is improved, the integrity and traceability of data are enhanced, and the safety and reliability of power grid dispatching are ensured.

[0007] To achieve the above object, the technical solution adopted by the present invention is:

[0008] A photovoltaic load control method based on fuzzy matching and topology dynamic optimization algorithm according to an embodiment of the present invention comprises the following steps:

[0009] S1. By pre-installing grid status data acquisition equipment, including sensors of distributed photovoltaic systems, the status data of each grid node and distributed photovoltaic system are acquired and recorded in real time;

[0010] S2. Construct a single-node discrete feature vector for the state data of each node of the power grid and the distributed photovoltaic system, assign different weight matrices to each feature, and construct the initial three-dimensional tensor of the power grid state. Decompose the initial three-dimensional tensor of the power grid state into sub-tensors using CP tensor decomposition, and finally fuse the decomposed sub-tensors to obtain the dynamic power grid state tensor DGST;

[0011] S3. Based on the historical data of single grid nodes and distributed photovoltaic systems, a benchmark grid state tensor is constructed as a reference standard;

[0012] S4. Eliminate abnormal dynamic grid state tensors in each single node of the grid and distributed photovoltaic system to ensure the accuracy and reliability of the data;

[0013] S5. Match multi-source grid data based on the data topology structure of grid nodes and distributed photovoltaic systems and fuzzy calculations, and identify and associate information from different data sources;

[0014] S6. Based on the matching results of multi-source power grid data, verify and optimize the multi-source power grid data, including distributed photovoltaic system data, to ensure the consistency and accuracy of the data;

[0015] S7. Define the grid data structure, store the verified and optimized multi-source grid data version, including the data of the distributed photovoltaic system, and record the generation time, version number, and modification record of the drawings, and save the grid status data section;

[0016] S8: Perform multi-copy management and simulation support based on all multi-source data verified and optimized in S6 and the data end surfaces saved in S7.

[0017] Optionally, S1 includes the following sub-steps:

[0018] S11. In this step, firstly, the data collection equipment is deployed at the key nodes in the power grid, including the installation of voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors and equipment status sensors in substations, distribution rooms and transmission lines, and voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors and equipment status sensors in distributed photovoltaic systems;

[0019] S12. In the above steps, data collection equipment is deployed at key nodes in the power grid. In this step, relevant data is collected, including voltage V, current I, power P, frequency f, phase angle θ, equipment status S, switch position K, load condition L, ambient temperature T, humidity H and number of faults F. In the distributed photovoltaic system, the voltage V of the distributed power source point is collected. DP , Distributed power point current I DP 、Distributed power point power P DP , Distributed power point frequency f DP 、Distributed power point phase angle θ DP , Distributed power supply point equipment status S DP .

[0020] Optionally, S2 includes the following sub-steps:

[0021] S21. Construct a single-node discrete feature vector for the state data of each node in the power grid through the real-time key feature vector formula:

[0022] X(t)=[V(t),I(t),P(t),f(t),θ(t),S(t),K(t),L(t),T(t),H(t),F(t),V DP (t),I DP (t),P DP (t),f DP (t),θ D X(t),S DP (t)];

[0023] Where t represents the time scale, X represents the discrete feature vector of a single node;

[0024] S22. Process the discrete characteristic data of a single node of the power grid, that is, first construct the initial three-dimensional tensor of the power grid state through the multi-time axis power grid tensor construction formula:

[0025] T(i,j,k)=X i (j)*W(k);

[0026] Among them, i represents different time points, j represents different features, k represents different weights, T represents the three-dimensional tensor of the initial power grid state, and W represents the weight matrix;

[0027] S23. In this step, the initial three-dimensional grid state tensor is decomposed into multiple sub-tensors of rank 3 based on the CP tensor decomposition formula to extract the main components and potential patterns. That is, the original tensor data usually contains a lot of redundant and noisy information. The main components can be extracted by decomposition, the data dimension can be reduced, and the noise can be removed. The CP tensor decomposition formula is as follows:

[0028] (λ r ,a r ,b r ,c r ) = Cp(T);

[0029] Among them, a r ,b r ,c r is the column vector of the factor matrix, λ r is the scalar weight, Cp is the CP tensor decomposition;

[0030] S24. In step S23, the initial three-dimensional grid state tensor is decomposed to reduce the dimension and noise of the data. In this step, a dynamic grid state tensor DGST is constructed based on the dynamic grid state fusion formula:

[0031]

[0032] Optionally, S3 includes the following sub-steps:

[0033] S31. Construct the benchmark power grid state tensor. First, based on historical data, statistical analysis is performed to obtain the mean of various characteristics of a single node in the power grid. The benchmark single-node discrete feature vector is constructed through the benchmark real-time key feature vector formula:

[0034]

[0035] Among them, X baseline represents the benchmark single-node discrete feature vector, μ v Represents the voltage mean of a single node in the power grid, μ I Represents the average current value of a single node in the power grid, μ P Represents the average power of a single node in the power grid, μ f Represents the frequency mean of historical data of a single node in the power grid, μ θ Represents the mean phase angle of the historical data of a single node in the power grid, μ S Represents the voltage mean of a single node in the power grid, μ K Represents the voltage mean of a single node in the power grid, μ LRepresents the voltage mean of a single node in the power grid, μ T Represents the voltage mean of a single node in the power grid, μ H Represents the voltage mean of a single node in the power grid, μ F represents the average number of failures of a single node in the power grid in historical data, represents the average voltage of distributed photovoltaics, represents the average value of distributed photovoltaic current, represents the average distributed photovoltaic power, represents the mean frequency of distributed photovoltaics, represents the average phase angle of distributed photovoltaics, Indicates the mean value of the status of distributed photovoltaic equipment;

[0036] S32, repeating steps S22, S23 and S24 with the reference single-node discrete feature vector obtained in step S31 to obtain a reference grid state tensor DGST baseline .

[0037] Optionally, S4 includes the following sub-steps:

[0038] S41. The difference between the dynamic grid state tensor and the reference grid tensor is characterized by a dynamic standard error formula:

[0039] Error = DGST-DGST baseline

[0040] Among them, Error is the difference tensor between the dynamic power grid state tensor and the reference power grid tensor;

[0041] S42. Calculate the mean of the difference vector using the difference tensor mean calculation formula:

[0042]

[0043] Among them, μ Δ Expressed as the mean of the difference tensor;

[0044] S43. Calculate the standard deviation of the difference vector using the difference tensor standard deviation calculation formula:

[0045]

[0046] Among them, σ Δ is the standard deviation of the difference vector;

[0047] S44. The threshold range is determined by using the anomaly detection threshold formula. If the difference tensor exceeds the threshold range, it is considered to be abnormal. The anomaly detection threshold formula is expressed as:

[0048]

[0049] The above anomaly detection threshold formula can be used to determine whether the dynamic power grid state vector is an outlier. When the dynamic power grid state vector at a certain time node is an outlier, it is directly removed.

[0050] Optionally, S5 includes the following sub-steps:

[0051] S51, construct the single node data collected in step S12 to represent it as a topological structure, first construct a node set Z based on the substation and the distribution room, and then construct an edge set E based on the transmission line, wherein the characteristic category and corresponding value of each sub-element in the node set and the edge set are recorded and stored, and finally the graph topology formula is used to construct the global topological structure data of the power grid:

[0052] G=(Z,E)

[0053] Wherein, G represents the global topological structure data of the power grid;

[0054] S52. Different from the past, when performing fuzzy matching of topological structures, only the similarity between discrete features of a single node or edge is calculated. Then, the original feature vector of a single node only contains a single feature of a single time point or a single data source, which cannot fully reflect the dynamic global changes of nodes and edges at different times and conditions. It is easy to ignore important information that changes over time in the power grid system, resulting in inaccurate matching results. In addition, the original feature vector is sensitive to noise and errors in the data and is easily affected by abnormal data. This application proposes a topological fusion similarity formula for calculating the similarity of heterogeneous data sources:

[0055]

[0056] Among them, ρ(X 1 ,X 2 ,DGST 1 ,DGST 2 ,DGST baseline 1 ,DGST baseline 2 ) represents the similarity between a single node or edge of two heterogeneous data sources, X 1 ,X 2 Represents the discrete feature vectors of a single node or edge of two heterogeneous data sources, DGST 1 ,DGST 2 Represents the dynamic grid state tensor of a single node or edge of two heterogeneous data sources, DGST baseline 1 ,DGST baseline 2A baseline grid state tensor representing a single node or edge of two heterogeneous data sources.

[0057] Optionally, S6 includes the following sub-steps:

[0058] S61, match all topological nodes of the two heterogeneous data sources, calculate similarity, and match the two with the highest similarity, that is, determine the two heterogeneous data sources G 1 , G 2 Sort by similarity, from G 1 Take the nodes in order, and then from G 2 The point with the highest similarity calculated value is selected for verification and optimization. First, local topology verification is performed, and the domain verification formula is used to compare whether the adjacent nodes and their connection relationships are consistent:

[0059] N(v i )={v j |A ij =0};

[0060] Among them, N(v i ) represents node v i The domain node set A ij Represents node v i With node v j Like the number of nodes;

[0061] The domain feature similarity calculation formula is used to calculate the feature similarity of nodes and edges in the neighborhood to verify the consistency of the neighborhood structure:

[0062]

[0063] Among them, ρ(N(v i ),N(v' i )) represents v i Neighborhood node set and v' i The similarity of the neighborhood node set, ρ(v j ,v k ) indicates v i The j node in the neighborhood node set and the representation v' i The similarity of k nodes in the neighborhood node set;

[0064] S62, further verify and check the global topology consistency to check whether the degree of the nodes and the paths between the nodes are consistent;

[0065] S63, verifying the local topology verification and the global topology consistency verification through steps S61 and S62, and re-adjusting the data source with low similarity between the dynamic power grid state tensor and the reference power grid state tensor when the local topology verification is inconsistent, so that G 1 , G2 The neighboring connection relationship of the direct nodes is consistent. When the global topology consistency verification is inconsistent, the position of the nodes in the data source with low similarity between the dynamic power grid state tensor and the benchmark power grid state tensor in the graph is also readjusted to ensure that G 1 , G 2 The global topology is consistent.

[0066] Optionally, S7 includes the following sub-steps:

[0067] S71. Define the power grid data structure {VersionID, Timestamp, ModificationLog}, version number VersionID, generation time Timestamp, modification record ModificationLog. After each power grid data update, add 1 to the current version number, record Timestamp and ModificationLog, and store each version of the drawing in the database to ensure the accessibility of historical versions.

[0068] S72. Save the grid status data section manually and automatically to ensure the integrity and traceability of the data. First, save manually to allow the dispatcher to manually save the current grid status as section data (CASE), and record the current real-time data, model information and operation log.

[0069] S73. This application proposes an automated end-face data preservation method, that is, a comprehensive evaluation index is calculated according to a trigger condition formula. When the comprehensive evaluation index is higher than a certain value, it is directly determined that an automatic trigger is required to save the power grid status data end face. The trigger condition formula is expressed as:

[0070]

[0071] Among them, CEI represents the comprehensive evaluation index, V nom ,I nom , P nom 、f nom They represent voltage rating, current rating, power rating and frequency rating respectively. bp(X) represents the probability of fault prediction for a single-node discrete feature vector X using a basic three-layer regression network. Tri(X) represents the judgment of all features within the single-node discrete feature vector X in combination with the physical constraints of the power grid operation. When there is a physical feature value that exceeds the physical constraint, it is recorded as 1.

[0072] When the CEI value is greater than 1, the section data is automatically saved, and distributed storage and blockchain methods are used to ensure the security and non-tamperability of power grid data.

[0073] Optionally, S8 includes the following sub-steps:

[0074] S81, the multi-source power grid data verified and optimized in S6 and the cross-section data in S7 are simultaneously saved and simulated according to the time axis and dimension, and different operations and fault responses are performed in multiple dimensions and under multiple conditions to simulate the multi-channel response of the power grid, and the copies of the power grid under different operations and fault responses are stored;

[0075] S82. Based on the multi-copy management of S81, simulation verification is performed based on the multi-channel response multi-copy data of the power grid.

[0076] First, the improved power flow equation is used to calculate the power grid flow and determine the distribution of current, voltage and power in the power system. The power flow equation is expressed as:

[0077]

[0078] Among them, P i and Q i are the active power and reactive power of the grid nodes respectively, M ij and N ij are the real and imaginary parts of the admittance between nodes i and j, respectively, θ i and θ j are the voltage phase angles at node i and node j, V i and V j are the voltages at point i and node j, dot(DGST i ,DGST i ) represents the point product operation of the dynamic grid state vector of node i.

[0079] The operation status of the power grid is determined based on the calculation results of the improved power flow equation;

[0080] S83. Obtain the current power grid status and dispatch plan from the replica data to ensure the accuracy and completeness of the input data. Then, perform operation simulation according to the dispatch plan to simulate the specific response and changes of the power grid during the dispatch operation. Finally, evaluate and analyze the simulation results to verify whether the dispatch strategy meets the expected goals, identify and solve possible problems, and ensure the safety and reliability of power grid dispatch;

[0081] S84. According to the preset fault conditions and multiple copies of data, set and simulate the occurrence of faults to observe the dynamic response of the power grid when the fault occurs. Then, reproduce the fault process through the simulation system and comprehensively analyze the impact of the fault on the power grid. Finally, evaluate the recovery capability of the power grid to ensure that the power grid can quickly and effectively restore normal operation after the fault occurs;

[0082] S85. Obtain the current grid state and protection device parameters from the copy data to ensure the accuracy and completeness of the input data. Then, perform protection simulation based on the grid state and protection device parameters to simulate the action of the protection device under different operating conditions. Finally, evaluate and analyze the simulation results to confirm whether the action of the protection device is correct, verify the effectiveness of the relay protection strategy, and ensure that the grid can be protected in a timely and effective manner when an abnormality occurs.

[0083] S86, obtain the current grid state and automatic device parameters from the replica data, and simulate the device according to the grid state and the automatic device parameters to simulate the action of the safety automatic device under multiple replica data. Finally, evaluate and analyze the simulation results to confirm whether the action of the device is correct and verify the effectiveness of the safety automatic device strategy.

[0084] A power grid dispatch verification system based on fuzzy matching and topology dynamic optimization, characterized by comprising the following modules:

[0085] Data acquisition module: used to install various sensors at key nodes of the power grid and the locations of distributed photovoltaic systems, including voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors, and equipment status sensors, to ensure real-time acquisition of operating data of the power grid and distributed photovoltaic systems. The data acquisition module is connected to the feature vector construction module to collect and transmit real-time data;

[0086] Feature vector construction and tensor decomposition module: construct single-node discrete feature vectors for the state data of each node of the power grid and the distributed photovoltaic system, and assign different weight matrices to construct the initial three-dimensional tensor of the power grid state. Use CP tensor decomposition to decompose the initial three-dimensional tensor of the power grid state into sub-tensors, and finally obtain the dynamic power grid state tensor. The feature vector construction and tensor decomposition module is connected with the reference state tensor construction module to generate dynamic power grid state data;

[0087] Benchmark state tensor construction module: Based on the historical data of single-node power grid and distributed photovoltaic system, the benchmark power grid state tensor is constructed as a reference standard. This module calculates the mean of each feature through statistical analysis of historical data and generates a benchmark single-node discrete feature vector. The benchmark state tensor construction module is connected with the outlier removal module to provide benchmark data for comparative analysis.

[0088] Outlier removal module: used to remove abnormal dynamic grid state tensors in each single node of the grid and distributed photovoltaic system to ensure the accuracy and reliability of the data. This module determines and removes abnormal data that exceeds the threshold range by calculating the dynamic standard error, the mean and standard deviation of the difference tensor. The outlier removal module is connected to the multi-source data matching module to provide cleaned data for matching analysis;

[0089] Multi-source data matching and verification optimization module: Based on the data topology structure and fuzzy calculation of power grid nodes and distributed photovoltaic systems, multi-source power grid data matching is performed, information from different data sources is identified and associated, and matching results are verified and optimized. This module builds a global topology structure, calculates the similarity between nodes and edges, performs data matching and verification, and ensures data accuracy and consistency. The multi-source data matching and verification optimization module is connected to the data structure definition and storage module to save the optimized data version;

[0090] Data structure definition and storage module: defines the grid data structure, stores and verifies the optimized multi-source grid data version, including the data of distributed photovoltaic systems, and records the generation time, version number, and modification record of the drawings. At the same time, this module saves the grid status data section to ensure the integrity and traceability of the data. The data structure definition and storage module is connected with the multi-copy management and simulation support module to provide historical data for simulation analysis.

[0091] Multi-copy management and simulation support module: Based on the verified and optimized multi-source data and the saved data sections, multi-copy management and simulation support are performed, including data simulation of distributed photovoltaic systems. This module ensures the stability and reliability of the power grid through power flow calculation, scheduling simulation, fault simulation, protection device simulation and automatic device simulation. The multi-copy management and simulation support module is used to evaluate the operating status of the power grid, optimize the scheduling strategy, and improve the overall efficiency and reliability of power grid scheduling.

[0092] The beneficial effects of the present invention are:

[0093] (1) This scheme uses CP tensor decomposition technology to construct and decompose high-dimensional feature vectors of grid state data and distributed photovoltaic system data to form a dynamic grid state tensor. This process can effectively remove redundant and noisy information in the data and improve the accuracy and efficiency of data processing. Through the fusion of dynamic tensors, the dynamic changes of the power grid and distributed photovoltaic system are reflected in real time, ensuring the timeliness and accuracy of scheduling decisions, especially the ability to quickly respond to changes in distributed photovoltaic power generation.

[0094] (2) This solution defines the grid data structure and version management mechanism, supports manual and automatic storage of grid status data sections, and ensures data security and immutability through distributed storage and blockchain technology. The multi-copy management and simulation functions enable the grid’s response under different operations and fault conditions to be fully simulated and verified, improving the reliability and security of scheduling. In particular, in the integration of distributed photovoltaic systems, it is possible to simulate the impact of photovoltaic systems on the grid and optimize scheduling strategies.

[0095] (3) The present invention proposes a topology fusion similarity formula to calculate the similarity of heterogeneous data sources, and combines the dynamic grid state tensor and the benchmark grid state tensor to accurately match and verify the optimization of multi-source data. This formula not only takes into account the traditional data of the power grid, but also integrates the data of the distributed photovoltaic system, improving the accuracy and consistency of data processing. Through this method, the overall operation status of the power grid can be more accurately reflected, including the real-time contribution and fluctuation of distributed photovoltaic power generation, thereby optimizing the dispatch and management of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 This is a flow chart of the photovoltaic load control method based on fuzzy matching and topology dynamic optimization proposed by the present invention; Figure 2 This is a network schematic diagram of the present invention. DETAILED DESCRIPTION

[0097] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments:

[0098] refer to Figure 1 , a photovoltaic load control method and system based on fuzzy matching and topology dynamic optimization includes the following steps:

[0099] S1. Obtain and record the status data of each node of the power grid in real time through the pre-set power grid status data acquisition equipment;

[0100] S2. Construct a single-node discrete feature vector for the state data of each node in the power grid, assign different weight matrices to each feature to construct an initial three-dimensional tensor of the power grid state, decompose the initial three-dimensional tensor of the power grid state into sub-tensors using CP tensor decomposition, and finally fuse the decomposed sub-tensors to obtain the dynamic power grid state tensor DGST;

[0101] S3. Construct a benchmark power grid state tensor based on the historical data of a single power grid node;

[0102] S4, eliminating abnormal dynamic grid state tensors in each single node of the grid;

[0103] S5. Matching multi-source power grid data based on the topological structure of power grid node data and fuzzy calculation;

[0104] S6. Verify and optimize the multi-source power grid data based on the multi-source power grid data matching results;

[0105] S7. Define the power grid data structure, store the verified and optimized multi-source power grid data version, record the generation time, version number, and modification record of the drawings, and save the power grid status data section;

[0106] S8: Perform multi-copy management and simulation support based on all multi-source data verified and optimized in S6 and the data end surfaces saved in S7.

[0107] In this implementation, S1 includes the following sub-steps:

[0108] S11. In this step, firstly, the data collection equipment is deployed at the key nodes in the power grid, including the installation of various sensors in substations, distribution rooms and transmission lines. The substation part, including the main substation and auxiliary substation, will deploy voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors and equipment status sensors. The distribution room part, especially the distribution room of the main load center, will be installed with the same sensor configuration to ensure real-time monitoring of power load and equipment status. Voltage and current sensors, environmental monitoring sensors and fault detection sensors are deployed at key locations of long-distance high-voltage transmission lines;

[0109] S12. In the above steps, data collection equipment is deployed at key nodes in the power grid. In this step, relevant data is collected, including voltage V, current I, power P, frequency f, phase angle θ, equipment status S, switch position K, load condition L, ambient temperature T, humidity H and number of faults F;

[0110] In this implementation, S2 includes the following sub-steps:

[0111] S21. Construct a single-node discrete feature vector for the state data of each node in the power grid through the real-time key feature vector formula:

[0112] X(t)=[V(t),I(t),P(t),f(t),θ(t),S(t),K(t),L(t),T(t),H(t),F(t)];

[0113] Where t represents the time scale, X represents the discrete feature vector of a single node;

[0114] S22. Process the discrete characteristic data of a single node of the power grid, that is, first construct the initial three-dimensional tensor of the power grid state through the multi-time axis power grid tensor construction formula:

[0115] T(i,j,k)=X i (j)*W(k);

[0116] Among them, i represents different time points, j represents different features, k represents different weights, T represents the three-dimensional tensor of the initial power grid state, and W represents the weight matrix;

[0117] S23. In this step, the initial three-dimensional grid state tensor is decomposed into multiple sub-tensors of rank 3 based on the CP tensor decomposition formula to extract the main components and potential patterns. That is, the original tensor data usually contains a lot of redundant and noisy information. The main components can be extracted by decomposition, the data dimension can be reduced, and the noise can be removed. The CP tensor decomposition formula is as follows:

[0118] (λ r ,a r ,b r ,c r ) = Cp(T);

[0119] Among them, a r ,b r ,c r is the column vector of the factor matrix, λ r is the scalar weight, Cp is the CP tensor decomposition;

[0120] S24. In step S23, the initial three-dimensional grid state tensor is decomposed to reduce the dimension and noise of the data. In this step, a dynamic grid state tensor DGST is constructed based on the dynamic grid state fusion formula:

[0121]

[0122] In this implementation, S3 includes the following sub-steps:

[0123] S31. Construct the benchmark power grid state tensor. First, based on historical data, statistical analysis is performed to obtain the mean of various characteristics of a single node in the power grid. The benchmark single-node discrete feature vector is constructed through the benchmark real-time key feature vector formula:

[0124] X baseline =[μ v ,μ I ,μ P ,μ f ,μ θ ,μ S ,μ K ,μ L ,μ T ,μ H ,μ F ];

[0125] Among them, X baseline represents the benchmark single-node discrete feature vector, μ v Represents the voltage mean of a single node in the power grid, μ I Represents the average current value of a single node in the power grid, μ P Represents the average power of a single node in the power grid, μ f Represents the frequency mean of historical data of a single node in the power grid, μ θRepresents the mean phase angle of the historical data of a single node in the power grid, μ S Represents the voltage mean of a single node in the power grid, μ K Represents the voltage mean of a single node in the power grid, μ L Represents the voltage mean of a single node in the power grid, μ T Represents the voltage mean of a single node in the power grid, μ H Represents the voltage mean of a single node in the power grid, μ F Indicates the average number of failures in the historical data of a single node in the power grid;

[0126] S32, repeating steps S22, S23 and S24 with the reference single-node discrete feature vector obtained in step S31 to obtain a reference grid state tensor DGST baseline ;

[0127] In this implementation, S4 includes the following sub-steps:

[0128] S41. The difference between the dynamic grid state tensor and the reference grid tensor is characterized by a dynamic standard error formula:

[0129] Error = DGST-DGST baseline

[0130] Among them, Error is the difference tensor between the dynamic power grid state tensor and the reference power grid tensor;

[0131] S42. Calculate the mean of the difference vector using the difference tensor mean calculation formula:

[0132]

[0133] Among them, μ Δ Expressed as the mean of the difference tensor;

[0134] S43. Calculate the standard deviation of the difference vector using the difference tensor standard deviation calculation formula:

[0135]

[0136] Among them, σ Δ is the standard deviation of the difference vector;

[0137] S44. The threshold range is determined by using the anomaly detection threshold formula. If the difference tensor exceeds the threshold range, it is considered to be abnormal. The anomaly detection threshold formula is expressed as:

[0138]

[0139] The above anomaly detection threshold formula can be used to determine whether the dynamic power grid state vector is an abnormal value. When the dynamic power grid state vector at a certain time node is an abnormal value, it is directly removed;

[0140] In this implementation, S5 includes the following sub-steps:

[0141] S51, construct the single node data collected in step S12 to represent it as a topological structure, first construct a node set Z based on the substation and the distribution room, and then construct an edge set E based on the transmission line, wherein the characteristic category and corresponding value of each sub-element in the node set and the edge set are recorded and stored, and finally the graph topology formula is used to construct the global topological structure data of the power grid:

[0142] G=(Z,E)

[0143] Wherein, G represents the global topological structure data of the power grid;

[0144] S52. Different from the past, when performing fuzzy matching of topological structures, only the similarity between discrete features of a single node or edge is calculated. Then, the original feature vector of a single node only contains a single feature of a single time point or a single data source, which cannot fully reflect the dynamic global changes of nodes and edges at different times and conditions. It is easy to ignore important information that changes over time in the power grid system, resulting in inaccurate matching results. In addition, the original feature vector is more sensitive to noise and errors in the data and is easily affected by abnormal data.

[0145] Therefore, based on the above, this application proposes a topological fusion similarity formula to calculate the similarity of heterogeneous data sources:

[0146]

[0147] Among them, ρ(X 1 ,X 2 ,DGST 1 ,DGST 2 ,DGST baseline 1 ,DGST baseline 2 ) represents the similarity between a single node or edge of two heterogeneous data sources, X 1 ,X 2 Represents the discrete feature vectors of a single node or edge of two heterogeneous data sources, DGST 1 ,DGST 2 Represents the dynamic grid state tensor of a single node or edge of two heterogeneous data sources, DGST baseline 1 ,DGST baseline 2A baseline grid state tensor representing a single node or edge of two heterogeneous data sources;

[0148] Based on the above, this application introduces the dynamic grid state tensor and the benchmark grid state tensor when calculating the similarity. The former can reflect the real-time multi-dimensional matching degree of two different data sources, and further matches the two through the benchmark grid state tensor. The benchmark grid state vector is calculated by integrating a large amount of historical data, so it represents the physical reality constraints of this grid node to a certain extent. Therefore, the historical similarity and physical constraints between the two heterogeneous data sources can be uniformly considered;

[0149] In this implementation, S6 includes the following sub-steps:

[0150] S61, match all topological nodes of the two heterogeneous data sources, calculate similarity, and match the two with the highest similarity, that is, determine the two heterogeneous data sources G 1 , G 2 Sort by similarity, from G 1 Take the nodes in order, and then from G 2 The point with the highest similarity calculated value is selected for verification and optimization. First, local topology verification is performed, and the domain verification formula is used to compare whether the adjacent nodes and their connection relationships are consistent:

[0151] N(v i )={v j |A ij =0};

[0152] Among them, N(v i ) represents node v i The domain node set A ij Represents node v i With node v j Like the number of nodes;

[0153] The domain feature similarity calculation formula is used to calculate the feature similarity of nodes and edges in the neighborhood to verify the consistency of the neighborhood structure:

[0154]

[0155] Among them, ρ(N(v i ),N(v' i )) represents v i Neighborhood node set and v' i The similarity of the neighborhood node set, ρ(v j ,v k ) indicates v i The j node in the neighborhood node set and the representation v' iThe similarity of k nodes in the neighborhood node set;

[0156] S62, further verify and check the global topology consistency to check whether the degree of the nodes and the paths between the nodes are consistent;

[0157] S63, verifying the local topology verification and the global topology consistency verification through steps S61 and S62, and re-adjusting the data source with low similarity between the dynamic power grid state tensor and the reference power grid state tensor when the local topology verification is inconsistent, so that G 1 , G 2 The neighboring connection relationship of the direct nodes is consistent. When the global topology consistency verification is inconsistent, the position of the nodes in the data source with low similarity between the dynamic power grid state tensor and the benchmark power grid state tensor in the graph is also readjusted to ensure that G 1 , G 2 The global topology is consistent;

[0158] In this implementation, S7 includes the following sub-steps:

[0159] S71. Define the power grid data structure {VersionID, Timestamp, ModificationLog}, version number VersionID, generation time Timestamp, modification record ModificationLog. After each power grid data update, add 1 to the current version number, record Timestamp and ModificationLog, and store each version of the drawing in the database to ensure the accessibility of historical versions.

[0160] S72. Save the grid status data section manually and automatically to ensure the integrity and traceability of the data. First, save manually to allow the dispatcher to manually save the current grid status as section data (CASE), and record the current real-time data, model information and operation log.

[0161] S73. This application proposes an automated end-face data preservation method, that is, a comprehensive evaluation index is calculated according to a trigger condition formula. When the comprehensive evaluation index is higher than a certain value, it is directly determined that an automatic trigger is required to save the power grid status data end face. The trigger condition formula is expressed as:

[0162]

[0163] Among them, CEI represents the comprehensive evaluation index, V nom ,I nom , P nom 、f nomThey represent voltage rating, current rating, power rating and frequency rating respectively. bp(X) represents the probability of fault prediction for a single-node discrete feature vector X using a basic three-layer regression network. Tri(X) represents the judgment of all features within the single-node discrete feature vector X in combination with the physical constraints of the power grid operation. When there is a physical feature value that exceeds the physical constraint, it is recorded as 1.

[0164] When the CEI value is greater than 1, the section data is automatically saved, and distributed storage and blockchain methods are used to ensure the security and non-tamperability of power grid data.

[0165] In this implementation, S8 includes the following sub-steps:

[0166] S81, the multi-source power grid data verified and optimized in S6 and the cross-section data in S7 are simultaneously saved and simulated according to the time axis and dimension, and different operations and fault responses are performed in multiple dimensions and under multiple conditions to simulate the multi-channel response of the power grid, and the copies of the power grid under different operations and fault responses are stored;

[0167] S82. Based on the multi-copy management of S81, simulation verification is performed based on the multi-channel response multi-copy data of the power grid.

[0168] First, the improved power flow equation is used to calculate the power grid flow and determine the distribution of current, voltage and power in the power system. The power flow equation is expressed as:

[0169]

[0170] Among them, P i and Q i are the active power and reactive power of the grid nodes respectively, M ij and N ij are the real and imaginary parts of the admittance between nodes i and j, respectively, θ i and θ j are the voltage phase angles at node i and node j, V i and V j are the voltages at point i and node j, dot(DGST i ,DGST i ) represents the point product operation of the dynamic grid state vector of node i.

[0171] The operation status of the power grid is determined based on the calculation results of the improved power flow equation;

[0172] S83. Obtain the current power grid status and dispatch plan from the replica data to ensure the accuracy and completeness of the input data. Then, perform operation simulation according to the dispatch plan to simulate the specific response and changes of the power grid during the dispatch operation. Finally, evaluate and analyze the simulation results to verify whether the dispatch strategy meets the expected goals, identify and solve possible problems, and ensure the safety and reliability of power grid dispatch;

[0173] S84. According to the preset fault conditions and multiple copies of data, set and simulate the occurrence of faults to observe the dynamic response of the power grid when the fault occurs. Then, reproduce the fault process through the simulation system and comprehensively analyze the impact of the fault on the power grid. Finally, evaluate the recovery capability of the power grid to ensure that the power grid can quickly and effectively restore normal operation after the fault occurs;

[0174] S85. Obtain the current grid state and protection device parameters from the copy data to ensure the accuracy and completeness of the input data. Then, perform protection simulation based on the grid state and protection device parameters to simulate the action of the protection device under different operating conditions. Finally, evaluate and analyze the simulation results to confirm whether the action of the protection device is correct, verify the effectiveness of the relay protection strategy, and ensure that the grid can be protected in a timely and effective manner when an abnormality occurs.

[0175] S86, obtain the current grid state and automatic device parameters from the replica data, and simulate the device according to the grid state and the automatic device parameters to simulate the action of the safety automatic device under multiple replica data. Finally, evaluate and analyze the simulation results to confirm whether the action of the device is correct and verify the effectiveness of the safety automatic device strategy.

[0176] So far, the technical route of the present invention has been described in detail through examples and in combination with the accompanying drawings, but it is easy for researchers in the field to understand that the protection scope of the present invention is obviously not limited to these specific implementations. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. Photovoltaic load control method based on fuzzy matching and topology dynamic optimization, Features ； S1. By pre-installing grid status data acquisition equipment, including sensors of distributed photovoltaic systems, the status data of each grid node and distributed photovoltaic system are acquired and recorded in real time; S2. Construct a single-node discrete feature vector for the state data of each node of the power grid and the distributed photovoltaic system, assign different weight matrices to each feature, construct an initial three-dimensional tensor of the power grid state, decompose the initial three-dimensional tensor of the power grid state into sub-tensors using canonical decomposition, and finally fuse the decomposed sub-tensors to obtain the dynamic power grid state tensor DGST; S3. Based on the historical data of single grid nodes and distributed photovoltaic systems, a benchmark grid state tensor is constructed as a reference standard; S4. Eliminate abnormal dynamic grid state tensors in each single node of the grid and distributed photovoltaic system to ensure the accuracy and reliability of the data; S5. Match multi-source grid data based on the data topology structure of grid nodes and distributed photovoltaic systems and fuzzy calculations, and identify and associate information from different data sources; S6. Based on the matching results of multi-source power grid data, verify and optimize the multi-source power grid data, including distributed photovoltaic system data, to ensure the consistency and accuracy of the data; S7. Define the grid data structure, store the verified and optimized multi-source grid data version, including the data of the distributed photovoltaic system, and record the generation time, version number, and modification record of the drawings, and save the grid status data section; S8: Perform multi-copy management and simulation support based on all multi-source data verified and optimized in S6 and the data end surfaces saved in S7.

2. According to claim 1, the photovoltaic load control method based on fuzzy matching and topology dynamic optimization, It is characterized in that The S1 comprises the following sub-steps: S11. In this step, firstly, the data collection equipment is deployed at the key nodes in the power grid, including the installation of voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors and equipment status sensors in substations, distribution rooms and transmission lines, and voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors and equipment status sensors in distributed photovoltaic systems; S12. In the above steps, data collection equipment is deployed at key nodes in the power grid. In this step, relevant data is collected, including voltage V, current I, power P, frequency f, phase angle θ, equipment status S, switch position K, load condition L, ambient temperature T, humidity H and number of faults F. In the distributed photovoltaic system, the voltage V of the distributed power source point is collected. DP , Distributed power point current I DP 、Distributed power point power P DP , Distributed power point frequency f DP 、Distributed power point phase angle θ DP , Distributed power supply point equipment status S DP .

3. According to claim 1, the photovoltaic load control method based on fuzzy matching and topology dynamic optimization, It is characterized in that The S2 comprises the following sub-steps: S21. Construct a single-node discrete feature vector for the state data of each node in the power grid through the real-time key feature vector formula: X(t)=[V(t), I(t), P(t), f(t), θ(t), S(t), K(t), L(t), T(t), H(t), F(t), VDP(t), I DP (t),P DP (t),f DP (t),θ DP (t),S DP (t)]; Where t represents the time scale, X represents the discrete feature vector of a single node; S22. Process the discrete characteristic data of a single node of the power grid, that is, first construct the initial three-dimensional tensor of the power grid state through the multi-time axis power grid tensor construction formula: T(i,j,k)=X i (j)*W(k); Among them, i represents different time points, j represents different features, k represents different weights, T represents the three-dimensional tensor of the initial power grid state, and W represents the weight matrix; S23. In this step, the initial three-dimensional grid state tensor is decomposed into multiple sub-tensors of rank 3 based on the CP tensor decomposition formula to extract the main components and potential patterns. That is, the original tensor data usually contains a lot of redundant and noisy information. The main components can be extracted by decomposition, the data dimension can be reduced, and the noise can be removed. The CP tensor decomposition formula is as follows: (λ r ,a r ,b r ,c r )=Cp(T); Among them, a r , b r , c r is the column vector of the factor matrix, λ r is the scalar weight, Cp is the CP tensor decomposition; S24. In step S23, the initial three-dimensional grid state tensor is decomposed to reduce the dimension and noise of the data. In this step, a dynamic grid state tensor DGST is constructed based on the dynamic grid state fusion formula:

4. The photovoltaic load control method based on fuzzy matching and topology dynamic optimization according to claim 1, It is characterized in that The S3 comprises the following sub-steps: S31. Construct the benchmark power grid state tensor. First, based on historical data, statistical analysis is performed to obtain the mean of various characteristics of a single node in the power grid. The benchmark single-node discrete feature vector is constructed through the benchmark real-time key feature vector formula: Among them, X baseline represents the benchmark single-node discrete feature vector, μ v Represents the voltage mean of a single node in the power grid, μ I Represents the average current value of a single node in the power grid, μ P Represents the average power of a single node in the power grid, μ f Represents the frequency mean of historical data of a single node in the power grid, μ θ Represents the mean phase angle of the historical data of a single node in the power grid, μ S Represents the voltage mean of a single node in the power grid, μ K Represents the voltage mean of a single node in the power grid, μ L Represents the voltage mean of a single node in the power grid, μ T Represents the voltage mean of a single node in the power grid, μ H Represents the voltage mean of a single node in the power grid, μ F represents the average number of failures of a single node in the power grid in historical data, represents the average voltage of distributed photovoltaics, represents the average value of distributed photovoltaic current, represents the average value of distributed photovoltaic power, represents the mean frequency of distributed photovoltaics, represents the average phase angle of distributed photovoltaics, Indicates the mean value of the status of distributed photovoltaic equipment; S32, repeating steps S22, S23 and S24 with the reference single-node discrete feature vector obtained in step S31 to obtain a reference grid state tensor DGST baseline .

5. The photovoltaic load control method based on fuzzy matching and topology dynamic optimization according to claim 1, It is characterized in that The S4 comprises the following sub-steps: S41. The difference between the dynamic grid state tensor and the reference grid tensor is characterized by a dynamic standard error formula: Error=DGST-DGST baseline Among them, Error is the difference tensor between the dynamic power grid state tensor and the reference power grid tensor; S42. Calculate the mean of the difference vector using the difference tensor mean calculation formula: Among them, μ Δ Expressed as the mean of the difference tensor; S43. Calculate the standard deviation of the difference vector using the difference tensor standard deviation calculation formula: Among them, σ Δ is the standard deviation of the difference vector, the i-th dynamic grid state tensor; S44. The threshold range is determined by using the anomaly detection threshold formula. If the difference tensor exceeds the threshold range, it is considered to be abnormal. The anomaly detection threshold formula is expressed as: The above anomaly detection threshold formula can be used to determine whether the dynamic power grid state vector is an outlier. When the dynamic power grid state vector at a certain time node is an outlier, it is directly removed.

6. The photovoltaic load control method based on fuzzy matching and topology dynamic optimization according to claim 1, It is characterized in that The S5 comprises the following sub-steps: S51, construct the single node data collected in step S12 to represent it as a topological structure, first construct a node set Z based on the substation and the distribution room, and then construct an edge set E based on the transmission line, wherein the characteristic category and corresponding value of each sub-element in the node set and the edge set are recorded and stored, and finally the graph topology formula is used to construct the global topological structure data of the power grid: G=(Z,E) Among them, G represents the global topological structure data of the power grid, and E represents the edge set of the transmission line; S52. Different from the past, when performing fuzzy matching of topological structures, only the similarity between discrete features of a single node or edge is calculated. Then, the original feature vector of a single node only contains a single feature of a single time point or a single data source, which cannot fully reflect the dynamic global changes of nodes and edges at different times and conditions. It is easy to ignore important information that changes over time in the power grid system, resulting in inaccurate matching results. In addition, the original feature vector is sensitive to noise and errors in the data and is easily affected by abnormal data. This application proposes a topological fusion similarity formula for calculating the similarity of heterogeneous data sources: Among them, ρ(X 1 ,X 2 ,DGST 1 ,DGST 2 ,DGST baseline 1 ,DGST baseline 2 ) represents the similarity between a single node or edge of two heterogeneous data sources, X 1 ,X 2 Represents the discrete feature vectors of a single node or edge of two heterogeneous data sources, DGST 1 ,DGST 2 Represents the dynamic grid state tensor of a single node or edge of two heterogeneous data sources, DGST baseline 1 ,DGST baseline 2 A baseline grid state tensor representing a single node or edge of two heterogeneous data sources.

7. The photovoltaic load control method based on fuzzy matching and topology dynamic optimization according to claim 1, It is characterized in that The S6 comprises the following sub-steps: S61, match all topological nodes of the two heterogeneous data sources, calculate similarity, and match the two with the highest similarity, that is, determine the two heterogeneous data sources G 1 , G 2 Sort by similarity, from G 1 Take the nodes in order, and then from G 2 The point with the highest similarity calculated value is selected for verification and optimization. First, local topology verification is performed, and the domain verification formula is used to compare whether the adjacent nodes and their connection relationships are consistent: N(v i )={v j |A ij =0}; Among them, N(v i ) represents node v i The domain node set A ij Represents node v i With node v j Like the number of nodes; The domain feature similarity calculation formula is used to calculate the feature similarity of nodes and edges in the neighborhood to verify the consistency of the neighborhood structure: Among them, ρ(N(v i ),N(v′ i )) represents v i Neighborhood node set and v′ i The similarity of the neighborhood node set, ρ(v j ,v k ) indicates v i The j node in the neighborhood node set and the representation v′ i The similarity of k nodes in the neighborhood node set; S62, further verify and check the global topology consistency to check whether the degree of the nodes and the paths between the nodes are consistent; S63, verifying the local topology verification and the global topology consistency verification through steps S61 and S62, and re-adjusting the data source with low similarity between the dynamic power grid state tensor and the reference power grid state tensor when the local topology verification is inconsistent, so that G 1 , G 2 The neighboring connection relationship of the direct nodes is consistent. When the global topology consistency verification is inconsistent, the position of the nodes in the data source with low similarity between the dynamic power grid state tensor and the benchmark power grid state tensor in the graph is also readjusted to ensure that G 1 , G 2 The global topology is consistent.

8. The photovoltaic load control method based on fuzzy matching and topology dynamic optimization according to claim 1, It is characterized in that The S7 comprises the following sub-steps: S71. Define the power grid data structure {VersionID, Timestamp, ModificationLog}, with version number VersionID, generation time Timestamp, and modification record ModificationLog. After each power grid data update, add 1 to the current version number, record Timestamp and ModificationLog, and store each version of the drawing in the database to ensure the accessibility of historical versions. S72. Save the power grid status data section manually and automatically to ensure the integrity and traceability of the data. First, manually save to allow the dispatcher to manually save the current power grid status as section data (CASE), and record the current real-time data, model information and operation log; S73. This application proposes an automated end-face data preservation method, that is, a comprehensive evaluation index is calculated according to a trigger condition formula. When the comprehensive evaluation index is higher than a certain value, it is directly determined that an automatic trigger is required to save the power grid status data end face. The trigger condition formula is expressed as: Among them, CEI represents the comprehensive evaluation index, V nom ,I nom , P nom 、f nom They represent voltage rating, current rating, power rating and frequency rating respectively. bp(X) represents the probability of fault prediction for a single-node discrete feature vector X using a basic three-layer regression network. Tri(X) represents the judgment of all features within the single-node discrete feature vector X in combination with the physical constraints of the power grid operation. When there is a physical feature value that exceeds the physical constraint, it is recorded as 1. When the CEI value is greater than 1, the section data is automatically saved, and distributed storage and blockchain methods are used to ensure the security and non-tamperability of power grid data.

9. The photovoltaic load control method based on fuzzy matching and topology dynamic optimization according to any one of claims 1 to 8, It is characterized in that The S8 comprises the following sub-steps: S81, the multi-source power grid data verified and optimized in S6 and the cross-section data in S7 are simultaneously saved and simulated according to the time axis and dimension, and different operations and fault responses are performed in multiple dimensions and under multiple conditions to simulate the multi-channel response of the power grid, and the copies of the power grid under different operations and fault responses are stored; S82. Based on the multi-copy management of S81, simulation verification is performed based on the multi-channel response multi-copy data of the power grid; First, the improved power flow equation is used to calculate the power grid flow and determine the distribution of current, voltage and power in the power system. The power flow equation is expressed as: Among them, P i and Q i are the active power and reactive power of the grid nodes respectively, M ij and N ij are the real and imaginary parts of the admittance between nodes i and j, respectively, θ i and θ j are the voltage phase angles at node i and node j, V i and V j are the voltages at point i and node j, dot(KGST i ,DGST i ) represents the point multiplication of the dynamic power grid state vector of node i; The operation status of the power grid is determined based on the calculation results of the improved power flow equation; S83, obtaining the current power grid status and dispatch plan from the copy data, ensuring the accuracy and completeness of the input data, then performing operation simulation according to the dispatch plan, simulating the specific response and changes of the power grid during the dispatch operation, and finally evaluating and analyzing the simulation results to verify whether the dispatch strategy meets the expected goals, identify and solve possible problems, and ensure the safety and reliability of power grid dispatch; S84. According to the preset fault conditions and multiple copies of data, set and simulate the occurrence of faults, observe the dynamic response of the power grid when the fault occurs, then reproduce the fault process through the simulation system, comprehensively analyze the impact of the fault on the power grid, and finally, evaluate the recovery capability of the power grid to ensure that the power grid can quickly and effectively restore normal operation after the fault occurs; S85, obtaining the current grid state and protection device parameters from the copy data, ensuring the accuracy and completeness of the input data, then performing protection simulation according to the grid state and protection device parameters, simulating the action of the protection device under different operating conditions, and finally evaluating and analyzing the simulation results to confirm whether the action of the protection device is correct, verifying the effectiveness of the relay protection strategy, and ensuring that the grid can be protected in a timely and effective manner when an abnormality occurs; S86. Obtain the current grid state and automatic device parameters from the replica data, and perform device simulation based on the grid state and automatic device parameters to simulate the action of the safety automatic device under multiple replica data. Finally, evaluate and analyze the simulation results to confirm whether the action of the device is correct and verify the effectiveness of the safety automatic device strategy.

10. A power grid dispatch verification system based on fuzzy matching and topology dynamic optimization, It is characterized in that Includes the following modules: Data acquisition module: used to install various sensors at key nodes of the power grid and the locations of distributed photovoltaic systems, including voltage sensors, current sensors, power meters, frequency meters, phase angle sensors, environmental sensors, and equipment status sensors, to ensure real-time acquisition of operating data of the power grid and distributed photovoltaic systems. The data acquisition module is connected to the feature vector construction module to collect and transmit real-time data; Feature vector construction and tensor decomposition module: construct single-node discrete feature vectors for the state data of each node of the power grid and the distributed photovoltaic system, and assign different weight matrices to construct the initial three-dimensional tensor of the power grid state. Use CP tensor decomposition to decompose the initial three-dimensional tensor of the power grid state into sub-tensors, and finally obtain the dynamic power grid state tensor. The feature vector construction and tensor decomposition module is connected with the reference state tensor construction module to generate dynamic power grid state data; Benchmark state tensor construction module: Based on the historical data of single-node power grid and distributed photovoltaic system, the benchmark power grid state tensor is constructed as a reference standard. This module calculates the mean of each feature through statistical analysis of historical data and generates a benchmark single-node discrete feature vector. The benchmark state tensor construction module is connected with the outlier removal module to provide benchmark data for comparative analysis. Outlier removal module: used to remove abnormal dynamic grid state tensors in each single node of the grid and distributed photovoltaic system to ensure the accuracy and reliability of the data. This module determines and removes abnormal data that exceeds the threshold range by calculating the dynamic standard error, the mean and standard deviation of the difference tensor. The outlier removal module is connected to the multi-source data matching module to provide cleaned data for matching analysis; Multi-source data matching and verification optimization module: Based on the data topology structure and fuzzy calculation of power grid nodes and distributed photovoltaic systems, multi-source power grid data matching is performed, information from different data sources is identified and associated, and matching results are verified and optimized. This module builds a global topology structure, calculates the similarity between nodes and edges, performs data matching and verification, and ensures data accuracy and consistency. The multi-source data matching and verification optimization module is connected to the data structure definition and storage module to save the optimized data version; Data structure definition and storage module: defines the grid data structure, stores and verifies the optimized multi-source grid data version, including the data of distributed photovoltaic systems, and records the generation time, version number, and modification record of the drawings. At the same time, this module saves the grid status data section to ensure the integrity and traceability of the data. The data structure definition and storage module is connected with the multi-copy management and simulation support module to provide historical data for simulation analysis. Multi-copy management and simulation support module: Based on the verified and optimized multi-source data and the saved data sections, multi-copy management and simulation support are performed, including data simulation of distributed photovoltaic systems. This module ensures the stability and reliability of the power grid through power flow calculation, scheduling simulation, fault simulation, protection device simulation and automatic device simulation. The multi-copy management and simulation support module is used to evaluate the operating status of the power grid, optimize the scheduling strategy, and improve the overall efficiency and reliability of power grid scheduling.

Images