Power grid operation optimization method and system for grid change resulting in change of transmission capacity

Abstract

The application relates to the field of information technology and discloses a power grid operation optimization method and system for changes in transmission capacity caused by changes in a grid structure, which obtains real-time operation data of a power grid to construct a grid structure data set, further extracts topological features from the grid structure data set to update a dynamic topological model of the grid structure, combines a power flow calculation method to calculate changes in transmission capacity, and further analyzes power distribution dynamic characteristics of each branch; a principal component analysis method is used to extract a main feature vector representing the operation state of the grid structure from the changes in transmission capacity and the power distribution dynamic characteristics, the matching degree of the feature vector and the branch power distribution is analyzed, a spatial optimization representation representing the operation state of the grid structure is obtained, and the spatial optimization representation is combined with operation stability constraints and economic targets of the power grid; and a dynamic programming algorithm is used to generate an operation optimization strategy to control the power grid to execute; the application can adapt to dynamic changes in the grid structure, realize real-time optimization and adjustment of the operation state of the power grid, and improve the stability of the operation of the power grid.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of information technology, in particular to a power grid operation optimization method and system caused by changes in the grid structure. BACKGROUND

[0002] Power system dispatching is the core of ensuring the safe and stable operation of the power grid. Its importance lies in optimizing resource allocation and operation mode to ensure efficient transmission and reliable supply of electricity. With the large-scale access of new energy and rapid growth of load demand, the grid structure frequently changes, and the transmission capacity changes significantly, which puts higher requirements on the adaptability and accuracy of the dispatching method. However, existing dispatching methods mostly rely on experience or static models, which have obvious limitations in dealing with changes in grid structure. The main performance is that the dynamic representation ability of the grid structure change is insufficient, it is difficult to accurately capture the complex influence of topology adjustment on transmission capacity, and there is a lack of mathematical description of real-time changes, which leads to the final dispatching scheme formed cannot fully predict the system state when facing complex grid adjustment, thereby affecting the stability and economy of the power grid operation.

[0003] Therefore, how to solve the problem that the existing dispatching method cannot effectively deal with the influence of the change of the grid structure on the transmission capacity, and then realize the real-time optimization and adjustment of the operation state of the power grid, has become a technical problem that the technical personnel in the field are eager to solve. SUMMARY

[0004] The present application provides a power grid operation optimization method and system caused by changes in the grid structure, which solves the problem that the existing dispatching method cannot effectively deal with the influence of the change of the grid structure on the transmission capacity, and then realizes the real-time optimization and adjustment of the operation state of the power grid.

[0005] To solve the above technical problems, the first aspect of the present application provides a power grid operation optimization method caused by changes in the grid structure, comprising:

[0006] Obtain real-time operation data of the power grid to construct a grid structure data set, and extract topological features from the grid structure data set using graph theory to obtain a topological feature matrix to update the dynamic topology model of the grid, generate a real-time topology adjustment model;

[0007] Based on the real-time topology adjustment model, the power flow calculation method is used to calculate the power distribution and transmission capacity of each branch in the power grid, to obtain the transmission capacity change quantization result to analyze the power distribution dynamic characteristics of each branch, and generate the grid power distribution dynamic characteristics;

[0008] A high-dimensional data processing set is constructed by the grid power distribution dynamic characteristics and the transmission capacity change quantitative results, a principal component analysis method is used to extract the main characteristic vector of the grid operation state, and a grid state feature set is obtained;

[0009] According to the grid state feature set, the matching degree of the main characteristic vector and the branch power distribution is analyzed to generate a spatial optimization representation of the grid operation state, and the operation stability constraint and the economic target of the power grid are combined, and a dynamic programming algorithm is used to generate an operation optimization strategy to control the power grid to execute.

[0010] The second aspect of the application provides a power grid operation optimization system caused by grid changes leading to transmission capacity changes, comprising:

[0011] A grid topology adjustment module is configured to obtain real-time operation data of the power grid to construct a grid structure data set, and extract topology features from the grid structure data set using a graph theory method to obtain a topology feature matrix to update a dynamic topology model of the grid, and generate a real-time topology adjustment model.

[0012] A transmission capacity quantitative module is configured to calculate the power distribution and transmission capacity of each branch in the power grid based on the real-time topology adjustment model using a power flow calculation method, obtain transmission capacity change quantitative results to analyze the power distribution dynamic characteristics of each branch, and generate grid power distribution dynamic characteristics.

[0013] An operation state determination module is configured to construct a high-dimensional data processing set by the grid power distribution dynamic characteristics and the transmission capacity change quantitative results, use a principal component analysis method to extract the main characteristic vector of the grid operation state, and obtain a grid state feature set.

[0014] An optimization strategy execution module is configured to analyze the matching degree of the main characteristic vector and the branch power distribution according to the grid state feature set to generate a spatial optimization representation of the grid operation state, combine the operation stability constraint and the economic target of the power grid, and use a dynamic programming algorithm to generate an operation optimization strategy to control the power grid to execute.

[0015] Compared with the prior art, the beneficial effects of the embodiments of the application are at least one of the following:

[0016] (1) By obtaining real-time operation data of the power grid, constructing a dynamic topology model, and analyzing power distribution dynamic characteristics, potential problems in power grid operation can be found in time, adjustments can be made in advance, and faults caused by unreasonable power distribution or insufficient transmission capacity can be avoided, improving the reliability and stability of power grid operation.

[0017] (2) The real-time topology adjustment model can be dynamically adjusted according to the real-time operation state of the power grid, so that the power grid can better adapt to the dynamic change of the grid structure, realize the real-time optimization adjustment of the power grid operation state, and improve the stability, flexibility and adaptability of the power grid operation;

[0018] (3) The main feature vector extracted based on the principal component analysis method and the space optimization representation provide a scientific basis for power grid operation decision-making, avoiding the uncertainty caused by subjective judgment and experience decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0020] Figure 1 is a flow chart of a power grid operation optimization method caused by grid structure change leading to change of transmission capacity provided by an embodiment of the present application;

[0021] Figure 2 is a structural diagram of a power grid operation optimization system caused by grid structure change leading to change of transmission capacity provided by an embodiment of the present application;

[0022] Reference signs:

[0023] Among them, 10, grid topology adjustment module; 20, transmission capacity quantification module; 30, operation state determination module; 40, optimization strategy execution module. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings and embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. The purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0025] In the description of the present application, the terms "first", "second", "third" and the like are only used for description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second", "third" and the like can explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "multiple" is two or more.

[0026] In the description of the present application, it is necessary to point out that, unless otherwise explicitly defined and limited, the terms "mounting", "connection", "connection" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the communication inside two elements. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the system or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0027] In the description of the present application, it is necessary to point out that, unless otherwise defined, all technical and scientific terms used by the present application are the same as the meanings generally understood by those skilled in the art. The terms used in the specification of the present application are only for describing specific embodiments, and are not intended to limit the present application. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0028] In an embodiment, as shown in Figure 1 The first aspect of the present application provides a power grid operation optimization method for changing the transmission capacity caused by the change of the grid, comprising:

[0029] S1, obtaining real-time operation data of the power grid to construct a grid structure data set, and extracting topological features from the grid structure data set by graph theory to obtain a topological feature matrix to update the dynamic topological model of the grid, and generating a real-time topological adjustment model;

[0030] In an embodiment, the real-time operation data of the power grid is obtained to construct a grid structure data set, comprising:

[0031] Real-time acquisition of node voltage value, branch current value and topological connection state in the power grid to construct an original data set and pre-process to obtain a cleaned data set;

[0032] The cleaned data set is constructed into a time series data set in time sequence to extract the dynamic features of the grid structure by sliding window method to obtain the grid structure data set representing the description of the grid structure change.

[0033] Specifically, the present application obtains the node voltage value, branch current value and topological connection state data in the power grid through sensors, smart meters, power grid automation systems (such as SCADA systems), power grid geographic information system (GIS) data, etc. in real time, and stores these data in a certain format and structure to form an original data set; the original data set is preprocessed by using data cleaning techniques, such as using interpolation method, mean filling method, etc. to fill in missing values; the abnormal values in the data are identified by statistical analysis methods (such as 3σ principle, box plot method, etc.) for correction or rejection; in order to reduce the noise interference in the data, moving average method, exponential smoothing method, etc. can be used for smoothing processing, etc. to obtain a cleaned data set; the cleaned data is arranged in time sequence, such as according to minute, hour or day time interval, to form a time series data set, each data point contains node voltage value, branch current value and topological connection state information, and corresponding time stamp, so as to analyze the change of the grid structure in different time scales.

[0034] According to the analysis requirements and data characteristics, the size and sliding step of the sliding window are determined; wherein the window size determines the amount of data contained in each analysis, and the sliding step determines the interval of window movement. For example, if it is desired to analyze the dynamic change of the grid structure within each hour, the window size can be set to 1 hour of data amount, and the sliding step is set to 15 minutes, so that more data segments can be obtained for analysis, and in each sliding window, the feature quantity capable of representing the dynamic characteristics of the grid structure is extracted: statistical characteristics, such as the average value, maximum value, minimum value, standard deviation, etc. of the node voltage and branch current, which can reflect the running state and fluctuation of the power grid in different time periods; correlation characteristics, the correlation coefficient between node voltages, between branch currents and between node voltages and branch currents is calculated by using correlation algorithms such as Pearson correlation coefficient, which is used to analyze the mutual influence relationship between different parts of the power grid; topological change characteristics, the number of topological connection state changes and change types (such as switch opening and closing operation) in the window are counted, to reflect the dynamic adjustment of the grid structure; the feature quantities extracted in each sliding window are combined to form a grid structure data set representing the description of the grid structure change, which can be used for subsequent power grid operation analysis, fault prediction and optimization decision, etc. For example, machine learning algorithms can be used to model and analyze the grid structure data set to predict the trend of future grid structure change, such as predicting that the power grid will add 5 branches within the next 24 hours based on historical data.

[0035] In addition, the cleaning data set can also be processed by using a graph algorithm to analyze the connection state of the power grid topology, to construct an adjacency matrix of the power grid, for example, the power grid contains 100 nodes and 150 branches at a certain time, the dimension of the adjacency matrix is 100*100, by comparing the adjacency matrices at different time points, the change of the grid structure can be identified as the dynamic characteristics of the grid structure, and in order to further describe the change of the grid structure, the average path length and the clustering coefficient of the network can be calculated by using a dynamic network analysis method as the dynamic characteristics of the grid structure, for example, the average path length of the network at a certain time is 3.2, and the clustering coefficient is 0.75, and the change of these indexes can reflect the stability and reliability of the power grid structure.

[0036] The power grid operation data is acquired in real time, various parameter changes of the power grid in the operation process can be captured in time, the actual operation condition of the power grid can be accurately understood, the original data set is preprocessed, the accuracy and integrity of the data can be ensured, the cleaned data set is constructed into a time sequence data set in time sequence, the dynamic characteristics of the grid structure are extracted by using a sliding window method, the change of the grid structure in different time periods can be effectively captured, potential problems and risks in the operation process of the power grid can be found in time, and the grid structure data set representing the change of the grid structure can provide an important basis for operation scheduling, planning and design and fault processing of the power grid.

[0037] In an embodiment, the topology characteristics are extracted from the grid structure data set by using a graph method to obtain a topology characteristic matrix, including:

[0038] Based on the grid structure data set, a depth-first search algorithm is used to traverse all nodes in the power grid to calculate the degree of each node, the weight of each branch is calculated by using Ohm's law, and the shortest path between each node is calculated by using a Floyd-Warshall algorithm to generate a connectivity matrix;

[0039] According to the degree of each node, the weight of each branch and the connectivity matrix, an initial topology characteristic matrix is constructed, and a principal component analysis method is used to extract main topology characteristics in the initial topology characteristic matrix;

[0040] The main topology characteristics are analyzed by using a K-means clustering algorithm, nodes and branches with similar topology characteristics are identified to construct node distribution groups, and the branches in the node distribution groups are optimized by using a minimum spanning tree algorithm to obtain an optimized branch connection scheme to update the initial topology characteristic matrix to obtain a final topology characteristic matrix.

[0041] Specifically, this invention, based on a network structure dataset, employs a depth-first search algorithm from graph theory to traverse the network nodes. Starting from a node in the network, it is marked as visited, and then all unvisited adjacent nodes of that node are visited sequentially. This process is repeated until all nodes have been visited. During the traversal, the number of adjacent nodes or branches of each node is counted as the degree of that node. For example, the degree of node E is 4, indicating that it is connected to 4 branches. Based on the current and voltage data of the branches, Ohm's law is used to calculate the weight of each branch. For example, the resistance value (or equivalent impedance) can be used directly as the branch weight, or the branch weight can be calculated according to the voltage drop requirement (such as using the voltage drop percentage as the weight index). The Floyd-Warshall algorithm is used to update the shortest path between nodes step by step through dynamic programming. The matrix is ​​updated through a triple loop until the shortest path of all node pairs is determined. Then, based on the shortest path calculation results between each node, a connectivity matrix is ​​constructed. For example, the shortest path length between node G and node H is 2, indicating that they are connected by two branches.

[0042] Features such as node degree, branch weight, and connectivity matrix are integrated into a matrix form, where rows represent nodes and columns represent features (degree, weight, shortest path, etc.) to form an initial topological feature matrix. Principal component analysis is then used to reduce the dimensionality of the initial topological feature matrix to extract the main eigenvectors and use them as the main topological features. For example, if the cumulative contribution rate of the first three principal components is 85%, it indicates that they can effectively represent the topological features, and these three principal components can be used as the main topological features.

[0043] The K-means clustering algorithm is used to perform cluster analysis on the extracted main topological features, identifying nodes and branches with similar topological features. For example, the clustering result is 5 classes, each containing 20 nodes and 30 branches. Node distribution groups are constructed based on the clustering results. The branches within the node distribution groups are optimized using the minimum spanning tree algorithm. A graph is constructed using branch weights as edge weights, and the Prim or Kruskal algorithm is applied to find the minimum spanning tree to optimize branch connections while maintaining connectivity, reducing redundant branches and lowering construction costs. This yields an optimized branch connection scheme, which is then integrated into the initial topological feature matrix. The node degree, branch weights, and other information are updated to form the final topological feature matrix.

[0044] This invention extracts power grid topology features from different perspectives using various algorithms, providing a comprehensive and detailed description of the power grid's topology and offering a rich information foundation for subsequent power grid analysis, optimization, and control. Principal component analysis (PCA) is used to extract key topology features, effectively reducing data dimensionality, removing redundant information, making the feature matrix more concise and efficient, reducing computational complexity, while retaining key information and improving the accuracy and efficiency of the analysis. The minimum spanning tree algorithm is used to optimize branches in the node distribution grouping, resulting in optimized branch connection schemes. This improves the power grid's topology, reduces operating costs, enhances reliability and stability, and achieves optimized power grid operation.

[0045] This invention collects basic information about the power grid, including node information (such as node number, type, voltage level, injected power, etc.) and branch information (such as nodes at both ends of the branch, impedance, rated capacity, etc.). Based on the connection relationship between nodes and branches, branch weights are used as matrix values ​​to generate an initial adjacency matrix. A graph structure is used to represent the dynamic topology model. In this model, nodes represent equipment in the power grid (such as substations, generators, loads, etc.), edges represent branches connecting the equipment, and the weights of the edges are the corresponding values ​​in the adjacency matrix. Then, based on the final obtained topology feature matrix, the degree change value of each node and the weight change value of each branch are calculated. The node degree change value is the current degree minus the previous degree, and the branch weight change value is the current weight minus the previous weight. If the node degree change value exceeds a preset node degree change threshold, or the branch weight change value exceeds a preset weight change threshold, the dynamic topology model of the network is recalculated using the weighted adjacency matrix to generate a real-time topology adjustment model. For example, when the degree of a node increases from 3 to 5 or the weight of a branch changes from 0.03 to 0.08, if the preset threshold is exceeded (e.g., degree change ±2, weight change ±0.05), the system automatically triggers an adjacency matrix update. Taking node J as an example, its degree change causes the corresponding row of non-zero elements in the adjacency matrix to be updated from [0,1,0,1,1] to [0,1,1,1,1,1]. At the same time, the weight of branch K is adjusted from 0.03 to 0.08, so the corresponding position value in its weighted adjacency matrix is ​​synchronously corrected, thereby updating the dynamic topology model of the network. The updated dynamic topology model is then subjected to connectivity analysis, vulnerability analysis, etc., to determine the connectivity and vulnerable links between nodes in the power grid. Based on the topology analysis results and the power grid's operational objectives (e.g., reliability, economy), a topology adjustment strategy is formulated. The adjustment strategy can be based on rules or optimization algorithms, which will not be detailed here. Finally, the adjustment strategy is integrated into the model to construct a real-time topology adjustment model. In addition, the model can be optimized and adjusted according to the actual situation during its operation to improve its accuracy and adaptability.

[0046] S2. Based on the real-time topology adjustment model, the power distribution and transmission capacity of each branch in the power grid are calculated using the power flow calculation method. The quantitative results of the transmission capacity change are obtained to analyze the dynamic characteristics of the power distribution of each branch and generate the dynamic characteristics of the power distribution of the grid.

[0047] In one embodiment, the step of calculating the power distribution and transmission capacity of each branch in the power grid based on the real-time topology adjustment model using the power flow calculation method to obtain the quantitative result of transmission capacity change includes:

[0048] Based on the real-time topology adjustment model, and combined with the injected power of each node and the impedance of each branch, the power flow is calculated using the Newton-Raphson method to obtain the power distribution data of each branch.

[0049] Based on the power distribution data of each branch, the transmission capacity change data of each branch is extracted to generate a transmission capacity quantification dataset. When the transmission capacity of a branch in the transmission capacity quantification dataset is lower than a preset transmission capacity threshold, the weight of each branch is adjusted to generate an optimized branch weight scheme.

[0050] The real-time topology adjustment model is optimized by the optimized branch weight scheme to obtain an optimized topology adjustment model for re-execution of power flow calculation until the transmission capacity of all branches in the final transmission capacity quantification dataset is not lower than the preset transmission capacity threshold. The final transmission capacity quantification dataset is then output as the transmission capacity change quantification result.

[0051] Specifically, this invention obtains the topology information of the power grid from a real-time topology adjustment model, including the connection relationships between nodes and branches. Simultaneously, it collects the injected power (power generated by generators or consumed by loads) of each node and the impedance parameters (resistance, reactance, etc.) of each branch. Based on this data, the Newton-Raphson method is used to calculate the power flow, obtaining the power distribution data for each branch: Initial values ​​of the voltage at each node are given (usually set to the rated voltage amplitude, with a phase angle of 0); according to the power flow equations, the elements of the Jacobian matrix are calculated. The Jacobian matrix is ​​the power flow equation applied to the node voltage... The partial derivative matrix of voltage amplitude and phase angle reflects the impact of node voltage changes on power balance. Based on the current node voltage value and injected power, the residual of the power flow equation (the difference between actual power and calculated power) is calculated. Then, the correction equation is solved using the Jacobian matrix to obtain the correction amount of node voltage amplitude and phase angle. The correction amount is added to the current node voltage value to obtain the new node voltage value. It is checked whether the residual of the new node voltage value is less than the set convergence accuracy. If the convergence condition is met, the iteration ends and the voltage and branch power distribution of each node are obtained; otherwise, the iterative calculation continues. For example, this invention uses the Newton-Raphson method to calculate power flow based on a real-time topology adjustment model and its updated adjacency matrix. The convergence accuracy is set to 1e-5. After 10 iterations, the node injection power distribution is obtained. For example, the injection power of node L changes from 1.2MW to 1.8MW. Combined with the branch impedance parameters (such as the impedance of branch M 0.05+j0.12), the power of each branch is calculated through the linearized power flow equation. It is found that the transmission capacity of branch N increases from 15MW to 18MW, while the transmission capacity of branch O decreases by 3MW due to the increase in weight.

[0052] Based on the branch power distribution data obtained from power flow calculations, and combined with the branch's rated power (or thermal stability limit power), the transmission capacity utilization rate (the ratio of actual power to rated power) of each branch is calculated. Simultaneously, the changes in branch transmission capacity at different times or under different operating modes are recorded, generating a quantitative dataset of transmission capacity. The transmission capacity utilization rate of each branch is compared with a preset transmission capacity threshold. If the transmission capacity utilization rate of a branch is lower than the threshold, it indicates that the branch has insufficient transmission capacity, and the weights of that branch and its related branches are adjusted accordingly. The branch weights can be... The weighting of branches is set based on factors such as impedance, length, and transmission capacity. For branches with insufficient transmission capacity, their weights can be appropriately reduced to encourage power flow to be transmitted through other branches. For branches with surplus transmission capacity, their weights can be appropriately increased to improve their utilization. Taking into account the overall operation of the power grid and optimization objectives (such as reducing power loss and improving power supply reliability), the weighting adjustments of each branch are combined to generate a reasonable optimized branch weighting scheme. This scheme should ensure that the adjusted branch weights can improve the power distribution of the power grid and increase the transmission capacity of the branches.

[0053] The optimized branch weighting scheme is applied to the real-time topology adjustment model to update the branch parameters in the model, such as modifying the branch weight attributes to ensure the model reflects the adjusted power grid structure. Using the updated real-time topology adjustment model, the Newton-Raphson method is used again to calculate power flow, resulting in new branch power distribution data and a quantified transmission capacity dataset. The new quantified transmission capacity dataset is then checked to see if any branches still have transmission capacities below a preset threshold. If so, the steps of adjusting branch weights and optimizing the model are repeated until the transmission capacity of all branches is above the preset threshold. The final quantified transmission capacity dataset is output as the quantified result of transmission capacity change. This result visually demonstrates the changes in transmission capacity of each branch under different operating conditions, providing important reference for power grid planning, operation, and maintenance.

[0054] This invention accurately obtains the power distribution and transmission capacity of each branch through power flow calculation, enabling a comprehensive and detailed understanding of the real-time operation of the power grid and providing accurate data for grid dispatching and management. By setting a preset transmission capacity threshold, when a branch's transmission capacity is found to be below this threshold, the branch weights are adjusted and the model optimized in a timely manner, effectively preventing grid faults caused by branch overload and improving grid security and stability. Based on the real-time topology adjustment model and power flow calculation results, the branch weights are rationally adjusted to optimize power distribution, enabling more efficient use of grid resources, reducing energy loss, and improving the economic efficiency of grid operation. This scheme can quickly respond to changes in grid topology and fluctuations in operating status. By continuously optimizing the model and adjusting branch weights, the grid can better adapt to different operating needs and emergencies, enhancing its adaptability and flexibility.

[0055] In one embodiment, obtaining the quantitative results of the transmission capacity change to analyze the dynamic characteristics of power distribution in each branch and generating dynamic characteristics of grid power distribution includes:

[0056] Based on the quantification results of the transmission capacity change, the power distribution data of each branch is extracted to calculate the correlation between the branch power and the rate of change of transmission capacity, and to generate a dynamic response feature description.

[0057] If the power change rate of a branch in the dynamic response feature description exceeds a preset power change rate threshold, the dynamic response feature description is adjusted by a weighted average method to obtain an adjusted dynamic response feature description.

[0058] A power allocation dynamic characteristic model is constructed based on the adjusted dynamic response characteristic description, and the power allocation dynamic characteristics of the grid structure are generated through the power allocation dynamic characteristic model.

[0059] Specifically, this invention extracts the power distribution data of each branch at each corresponding time based on the quantitative results of the change in transmission capacity. For each branch, it calculates its power change rate and transmission capacity change rate. The power change rate can be obtained by calculating the ratio of the power difference between adjacent time moments to the time interval, and the transmission capacity change rate is obtained in the same way. Correlation analysis methods (such as Pearson correlation coefficient) are used to calculate the correlation between the power change rate and the transmission capacity change rate of each branch, generating a dynamic response feature description, which is presented in the form of branch number, correlation coefficient, power change rate, etc.

[0060] Furthermore, by comparing the rate of change of each branch, a characteristic description reflecting the dynamic response speed of the power distribution of the grid can be generated, i.e., a dynamic response characteristic description. For example, if the power of branch A increases from 10MW to 12MW (a change rate of 20%), while the power of branch B decreases from 8MW to 7MW (a change rate of 12.5%), by comparing the rate of change of each branch, a characteristic description reflecting the dynamic response speed of the power distribution of the grid can be generated, i.e., the response speed of branch A is 0.2MW / s, and that of branch B is 0.125MW / s. Further, by combining branch impedance and node injected power, the power change curve can be fitted using the least squares method, and the dynamic characteristics of the power distribution of the grid, i.e., the dynamic response characteristic description, can be obtained through the correlation of power change or the response speed of power change.

[0061] The power change rate of each branch in the dynamic response feature description is compared with a preset power change rate threshold. For branches whose power change rate exceeds the threshold, a weighted average method is used for adjustment. Specifically, factors such as the historical data of the branch and the power change of adjacent branches can be comprehensively considered to assign an adjustment weight to the power change rate of the branch. Then, the adjusted power change rate is calculated by weighted averaging, and the dynamic response feature description is updated to obtain the adjusted dynamic response feature description. For example, assuming that the power change rate of branch 2 is 1.2, exceeding the preset threshold ±0.5, and the power change rates of adjacent branches 1 and 3 are 0.3 and 0.4 respectively, and the average power change rate of branch 2 in historical data is 0.6, the historical weights of these three adjacent branches are 0.2, 0.3, and 0.5 respectively. Then, the power change rate of branch 2 is calculated to be 0.5 using the weighted average method.

[0062] Using adjusted dynamic response characteristics as input, and grid topology information such as node and branch connections and branch impedance parameters, along with external disturbance information such as sudden load increases or decreases and generator output changes, a model reflecting the dynamic response speed of grid power distribution is constructed. This model can be a mathematical model, built using methods such as time series analysis, system dynamics, and neural network algorithms. It utilizes adjusted power change rate data to establish a dynamic response model for power distribution in each branch, predicting power distribution under different grid structures. Historical data is used to validate the model, ensuring its accurate reflection of the dynamic response speed of grid power distribution. Comparison with actual operating data evaluates the model's accuracy and reliability. Finally, the dynamic response speed characteristics of grid power distribution are output from the power distribution dynamic characteristic model as the dynamic features of grid power distribution, providing a basis for optimal grid scheduling.

[0063] This invention, by quantifying changes in transmission capacity and analyzing the dynamic characteristics of power distribution, provides a deeper understanding of the time-varying patterns of power distribution in various branches of the power grid. This helps to promptly identify unreasonable power distribution or potential operational risks, providing strong support for optimized grid scheduling and safe and stable operation. By calculating the correlation between branch power and the rate of change of transmission capacity, a dynamic response characteristic description is generated, accurately reflecting the response of each branch to changes in transmission capacity. This facilitates the identification of branches with abnormal responses, providing a clear direction for subsequent adjustments and optimizations. When the rate of change of power in a branch exceeds a preset threshold, a weighted average method is used to adjust the dynamic response characteristic description, making the adjusted description more consistent with the actual operation of the power grid. This improves the adaptability and accuracy of the constructed dynamic characteristic model of power distribution, thus more accurately reflecting the dynamic response speed of grid power distribution. The dynamic characteristics of grid power distribution generated based on the constructed dynamic characteristic model provide grid operators with intuitive and comprehensive decision-making basis, helping to formulate more scientific and reasonable operating strategies and improve the operating efficiency and reliability of the power grid.

[0064] S3. Construct a high-dimensional data processing set by using the dynamic characteristics of the power distribution of the grid structure and the quantitative results of the change in the transmission capacity, and use principal component analysis to extract the main feature vectors of the grid structure's operating state to obtain the grid structure state feature set;

[0065] Specifically, this invention standardizes and integrates the dynamic characteristics of power allocation in the grid structure with the quantitative results of changes in transmission capacity to form a high-dimensional data processing set. For example, the power allocation characteristics and the quantitative results of changes in transmission capacity can be added to the dataset as feature dimensions.

[0066] In one embodiment, the extraction of the main feature vectors of the grid structure's operating state using principal component analysis to obtain the grid structure's state feature set includes:

[0067] The high-dimensional data processing set is processed by principal component analysis to extract the main feature vectors characterizing the operation state of the grid structure, and a dynamic characterization of the operation state of the grid structure is generated based on the main feature vectors as the initial grid structure state feature set.

[0068] If there are major feature vectors in the initial grid state feature set with a variance less than a preset variance threshold, then the initial grid state feature set is updated by linear interpolation to obtain the grid state optimized feature set.

[0069] Based on the aforementioned network state optimization feature set, a dynamic characterization model of network operation state is constructed to generate a dynamic characterization vector of network topology. Based on the dynamic characterization vector, the dynamic interaction relationship between branch power and network topology is analyzed to obtain a stability feature description characterizing network power allocation.

[0070] According to the stability feature description, periodic features of the change in transmission capacity are extracted from the high-dimensional data processing set to generate a periodic feature dataset. When there is data in the periodic feature dataset whose change period exceeds a preset change period threshold, the periodic features are decomposed by Fourier transform to obtain a decomposed periodic feature set.

[0071] The space frame state optimization feature set is updated by decomposing the periodic feature set to obtain the space frame state target feature set as the output space frame state feature set.

[0072] Specifically, this invention processes high-dimensional data sets using principal component analysis. By calculating the covariance matrix, eigenvalues, and eigenvectors, it extracts the main eigenvectors characterizing the operating state of the grid structure. These eigenvectors are then used as dynamic representations of the grid structure's operating state to construct an initial grid structure state feature set. For example, by calculating the covariance matrix, the first three principal components are extracted, with contribution rates of 45%, 30%, and 15%, respectively, yielding a dynamic representation of the grid structure's operating state. The first principal component primarily reflects branch power changes, the second principal component reflects dynamic response speed, and the third principal component reflects dynamic characteristics. These are used as the main eigenvectors and integrated to generate the initial grid structure state feature set.

[0073] Calculate the variance of each major feature vector in the initial grid structure state feature set. If there is a major feature vector with a variance less than a preset variance threshold, it indicates that the feature has a weak ability to represent the grid structure's operating state. Then, use linear interpolation to update the initial grid structure state feature set based on the mean or median of its adjacent feature vectors to improve the quality of the feature set and obtain the optimized grid structure state feature set.

[0074] Based on the network structure state optimization feature set, a dynamic representation model of the network structure's operating state is constructed using methods such as system dynamics and neural networks. For example, when using the system dynamics method, key variables in the system are first identified (such as features in the network structure state optimization feature set, power of each branch, transmission capacity, load, etc.) as training data, and the causal relationships between them are analyzed to establish a stock flow diagram. Then, simulation software (such as Vensim) is used to solve the model, simulating the dynamic changes of the system. The network structure state optimization feature set is input into the constructed dynamic representation model of the network structure's operating state, and the dynamic representation vector of the network structure topology is output. Based on the dynamic representation vector, the dynamic interaction relationship between branch power and network structure topology is analyzed. For example, when the power of a certain branch suddenly increases, the power response of other branches in the network structure topology is observed, and how this response affects the power distribution stability of the entire network structure is analyzed. Based on the analysis results of the dynamic interaction relationship, a stability feature description is generated. This stability feature description is characterized as the stability index of network structure power distribution (such as power fluctuation range, response time, etc.) under specific operating conditions, and the relationship between these indexes and the network structure topology.

[0075] Based on the stability characteristics, periodic features of power fluctuations and corresponding time-related changes in transmission capacity are extracted from the high-dimensional data processing set. For example, time series analysis methods (such as Fourier transform) can be used to extract the periodic variation patterns of transmission capacity on the corresponding time scale, generating a periodic feature dataset. If the periodic feature dataset contains data with a variation period exceeding a preset threshold, these periodic features are decomposed using Fourier transform to obtain components of different frequencies and their corresponding amplitude and phase information, forming a decomposed periodic feature set. For example, complex periodic changes can be decomposed into sine or cosine wave components of different frequencies to more deeply analyze the intrinsic laws governing changes in transmission capacity.

[0076] The periodic features obtained from the decomposition are added as new dimensions to the grid state optimization feature set for updating. The updated grid state target feature set is output as the grid state feature set, which can more comprehensively reflect the operating status of the grid. At the same time, based on the information in the target feature set, the power distribution stability of the grid under different structures can be analyzed, thereby providing data support for power grid planning.

[0077] This invention effectively reduces the dimensionality of data by constructing a high-dimensional data processing set and employing principal component analysis, extracting the main feature vectors of the grid's operating state, removing redundant information, and highlighting key features. Based on the optimized feature set, a dynamic representation model is constructed, and the dynamic interaction between branch power and grid topology is analyzed, enabling a deeper understanding of the mutual influence between power distribution and topology in the power grid. Furthermore, by extracting periodic features of transmission capacity changes from the high-dimensional data processing set and performing Fourier transform decomposition on data exceeding a preset period threshold, the changing patterns of periodic features can be analyzed in more detail, further enriching the understanding of the grid's operating state, capturing more complex and subtle operating characteristics, and ultimately resulting in a more comprehensive and accurate reflection of the grid's operating state and power distribution characteristics.

[0078] In one embodiment, after extracting the main feature vectors of the grid structure's operating state using principal component analysis to obtain the grid structure's state feature set, the method further includes:

[0079] Calculate the variance contribution rate of each of the main feature vectors in the set of state features of the space frame;

[0080] When it is determined that the variance contribution rate of the main feature vector in the set of network state features is lower than the preset contribution threshold, a deep learning algorithm is used to perform nonlinear dimensionality reduction on the high-dimensional data processing set to obtain a dimensionality-reduced data processing set.

[0081] The dimensionality-reduced data set is processed using a convolutional neural network to extract features, resulting in the final network structure state feature set for output.

[0082] Specifically, this invention calculates the variance contribution rate of each major feature vector in the power grid state feature set. If the variance contribution rate of any feature vector is lower than a preset contribution threshold (e.g., 10%), or if the stability of the feature vector does not match the rate of change of the transport capacity (e.g., the fluctuation amplitude of the feature vector exceeds 5%), a deep learning algorithm is used to perform nonlinear dimensionality reduction on the high-dimensional data. An autoencoder model can be used, taking high-dimensional data (50 dimensions) containing the power grid state feature set and parameters such as branch power, rate of change, and dynamic characteristics as input. The data is compressed to 10 dimensions through the encoding layer of the model, and then the data is restored through the decoding layer of the model, outputting a dimensionality-reduced data processing set. In the training process of the autoencoder model, mean squared error is used as the loss function, so that the final optimized power grid state feature set can more accurately reflect the operating status. Furthermore, a convolutional neural network (CNN) is used to extract features from the dimensionality-reduced data processing set. Local features are extracted through convolutional layers and pooling layers to obtain the final power grid state feature set for output.

[0083] This invention calculates the variance contribution rate of the main feature vectors and uses a deep learning algorithm for nonlinear dimensionality reduction when the variance contribution rate is below a preset threshold. This can compensate for the shortcomings of PCA in processing nonlinear data and more comprehensively mine the information in the data. By automatically learning complex features in the data through deep learning algorithms and using CNN to extract features from the dimensionality-reduced data, the accuracy and effectiveness of feature extraction can be further improved, resulting in a feature set that better reflects the essence of the network structure's operating state.

[0084] S4. Based on the network state feature set, analyze the matching degree between the main feature vector and the branch power distribution to generate a spatial optimization representation of the network operation state and combine it with the operation stability constraints and economic objectives of the power grid. Use dynamic programming algorithm to generate operation optimization strategies to control the execution of the power grid.

[0085] In one embodiment, the step of analyzing the matching degree between the main feature vectors and the branch power distribution based on the grid state feature set to generate a spatial optimization representation of the grid operating state includes:

[0086] The spatial distribution characteristics of each main feature vector in the state feature set of the grid structure are calculated using the kernel density estimation method to obtain the feature distribution characterization vector. The probability density function of the power distribution of each branch is then calculated and matched with the feature distribution characterization vector using KL divergence to obtain the matching result.

[0087] The space where the matching results meet the matching criteria is taken as the target region, and a network topology model is created by using a graph convolutional network. The node voltage and branch impedance of the target region are input into the network topology model for processing, and node embedding vectors are output.

[0088] The node embedding vectors are spatially partitioned using the DBSCAN clustering algorithm to obtain several highly coordinated regions. The random forest regression method is then used to establish the mapping relationship between the main feature vectors in the target region and each of the highly coordinated regions to generate a spatially optimized representation matrix.

[0089] The dimensions of the spatial optimization representation matrix are compressed and missing values ​​are filled by bilinear interpolation to generate a spatial coordination heatmap as a spatial optimization representation output of the grid structure's operating status.

[0090] Specifically, this invention uses kernel density estimation (KDE) to calculate the spatial distribution characteristics of each major feature vector in the network state feature set, obtaining a feature distribution representation vector. For the power data of each branch, the same kernel density estimation method is used to calculate its probability density function. Then, the KL divergence between the feature distribution representation vector and the probability density function of the branch power distribution is calculated to obtain the matching result. The smaller the KL divergence value, the higher the degree of matching.

[0091] Using a preset KL divergence threshold of 0.03 as the matching standard, when the matching result meets the matching standard, the space where the branch is located is taken as the target area, indicating that the spatial distribution of the two is highly consistent. The power grid topology of the target area is modeled as a graph structure, where nodes represent nodes in the power grid (such as substations, buses, etc.) and edges represent branches. The node voltage and branch impedance of the target area are used as input features to construct an adjacency matrix and set the hidden layer dimension to 16. The ReLU activation function is used to extract spatial correlation features. The node features are aggregated and updated through multi-layer graph convolution operation of the graph convolutional network model, and the node embedding vectors are output. These vectors contain the spatial feature information of the nodes.

[0092] Using node embedding vectors as input data, the DBSCAN algorithm (neighborhood radius ε=0.5, minimum sample size=10) is run to cluster the node embedding vectors, outputting the cluster label to which each node belongs. Based on the cluster label, spatially adjacent and feature-similar nodes are grouped into the same high-coordination region, resulting in several high-coordination regions where the standard deviation of branch power fluctuation is less than 2%. Using the main feature vectors in the target region as input features and the high-coordination regions to which nodes belong as output features, a random forest regression model (number of trees=100, maximum depth=8) is constructed. By training the model, the intrinsic relationship between the main feature vectors and the high-coordination regions is learned. The trained model is used to predict all main feature vectors, obtaining the probability distribution of the high-coordination regions corresponding to each main feature vector, and then generating a spatially optimized representation matrix. The elements in the matrix represent the mapping relationship between feature vectors and high-coordination regions, which can also be understood as the degree of correlation.

[0093] The PCA algorithm is used to compress the dimension of the spatial optimization representation matrix. The covariance matrix of the spatial optimization representation matrix is ​​calculated, and the eigenvalues ​​and eigenvectors of the covariance matrix are obtained. The eigenvectors corresponding to the first few largest eigenvalues ​​are selected as the projection matrix to project the spatial optimization representation matrix into a low-dimensional space, resulting in a compressed matrix with a dimension of 8×8. Bilinear interpolation is then performed on the compressed matrix to fill missing values. A visualization tool (such as Matplotlib) is used to convert the filled matrix into a spatial compatibility heatmap. The color intensity represents the level of spatial compatibility, and this heatmap is used as the spatial optimization representation output of the grid structure's operating status.

[0094] This invention calculates the spatial distribution characteristics of feature vectors using kernel density estimation and combines it with KL divergence to match branch power distribution, enabling more accurate quantification of the matching degree between feature vectors and branch power distribution, thereby identifying areas with poor spatial coordination. It employs Graph Convolutional Networks (GCNs) to model the network topology of the target area, fully utilizing the topological information of the power grid to capture spatial relationships between nodes and extract more effective node embedding vectors, providing a foundation for subsequent spatial partitioning and coordination analysis. The DBSCAN clustering algorithm is used to spatially partition the node embedding vectors, automatically identifying highly coordinated areas. A mapping relationship between feature vectors and highly coordinated areas is established using random forest regression, generating a spatial optimization representation matrix that provides a scientific basis for power grid operation optimization. By compressing the dimension of the spatial optimization representation matrix and using bilinear interpolation to fill missing values, a spatial coordination heatmap is generated, which visually displays the spatial coordination status of the power grid, facilitating operators to quickly locate problem areas and take corresponding measures.

[0095] In one embodiment, the generation of a spatial optimization representation of the grid's operating state, combined with the grid's operational stability constraints and economic objectives, employs a dynamic programming algorithm to generate an operational optimization strategy, including:

[0096] Principal component analysis is used to reduce the dimensionality of the spatial optimization representation to obtain a dimensionality-reduced feature vector. When the dimension of the dimensionality-reduced feature vector is lower than a preset dimension threshold, the dimension of the dimensionality-reduced feature vector is expanded by Gaussian process regression to obtain an expanded feature vector.

[0097] Based on the extended feature vector, the dynamic response characteristics of the grid structure are determined, and the economic objective is to minimize the operating cost. The operating stability constraints are the power flow balance equation and the N-1 safety check. An operating optimization strategy is generated through a dynamic programming algorithm.

[0098] Specifically, this invention employs principal component analysis to reduce the dimensionality of the spatial optimization representation, obtaining dimensionality-reduced feature vectors. When the dimensionality of the dimensionality-reduced feature vectors is lower than a preset dimensionality threshold, a Gaussian process model is constructed using a squared exponential kernel as the kernel function. The dimensionality-reduced feature vectors are used as input data, and the corresponding known dimensions in the original spatial optimization representation are used as output data. A Gaussian process regression model is trained to learn the mapping relationship between input and output. Using the trained Gaussian process regression model, the dimensionality-reduced feature vectors are predicted to obtain estimates of the missing dimensions. These estimates are then combined with the original dimensionality-reduced feature vectors to obtain extended feature vectors.

[0099] The various feature components in the extended feature vector are analyzed to calculate their statistical characteristics (such as mean, variance, maximum, minimum, etc.) and the correlation between them. Based on the results of the feature analysis, combined with the knowledge and experience of power grid operation, the variation law of the feature vector over time is analyzed to identify the feature components closely related to the dynamic behavior of the grid, thereby describing the dynamic changes of the grid state as dynamic response features of the grid state, such as voltage fluctuation features and power change features. These features can be represented as time series or vector form to describe the dynamic response of the grid under different operating conditions.

[0100] The objective is to minimize operating costs, which can include generation costs, transmission loss costs, and equipment maintenance costs. The generation cost coefficient is set at 0.8 yuan / kWh, and the transmission loss cost coefficient at 1.2 yuan / kWh. These costs are quantified to construct an operating cost function, which serves as the objective function of the dynamic programming algorithm. The power flow balance equation is a fundamental constraint on power grid operation, ensuring the balance of active and reactive power at each node. This invention uses the power flow balance equation as an operational stability constraint to ensure that the optimized operating strategy meets the power balance requirements of the power grid. N-1 safety verification is an important criterion for safe power grid operation, requiring that any branch or device in the power grid... Even when out of operation, the power grid can still maintain safe and stable operation. This invention uses N-1 safety verification as an operational stability constraint to evaluate the safety of the optimization strategy. Based on the dynamic response characteristics, constraints, and objectives of the grid, a dynamic programming algorithm is used to establish the state transition equation. The state variables are defined as node voltage amplitude deviation (allowable range of ±5%) and branch power over-limit probability (threshold set to 10%). The action variables are generator output adjustment (step size 10MW) and the number of capacitor switching groups (2Mvar per group). The value iteration method is used to solve the problem, and the discount factor is set to 0.9. The convergence threshold is 1e-4. After 32 iterations, the operation optimization strategy is obtained to control the execution of the power grid. For example, during peak load periods in a certain region, the algorithm output increases the output of Unit 3 by 80MW and simultaneously activates two sets of capacitors, raising the power margin of critical sections from 8% to 15%. For scenarios involving changes in the network topology, the state space is dynamically updated based on sensitivity analysis. When a 500kV line is disconnected, 12 nodes with a calculated impedance matrix change rate exceeding 20% ​​are included in key monitoring. After recalculation, the output of four units is adjusted by a total of 150MW. The final dispatch instructions are issued via the OPC-UA protocol, containing 104 adjustment parameters for 17 control nodes. After execution, the measured network loss is reduced by 3.7%, and the voltage qualification rate is increased to 99.2%.

[0101] This invention employs principal component analysis to reduce the dimensionality of the spatial optimization representation, removing redundant information, reducing data dimensionality, and improving the computational efficiency of subsequent processing. Simultaneously, it retains key feature information to ensure data quality. When the dimensionality is too low after dimensionality reduction, Gaussian process regression is used to expand the dimensionality, compensating for any key information that may be lost during the dimensionality reduction process, making the feature vectors more comprehensively reflect the grid state. Based on the expanded feature vectors, the dynamic response characteristics of the grid state are determined, enabling more accurate capture of the dynamic changes of the grid under different operating conditions, providing a more reliable foundation for subsequent operational optimization. With minimum operating cost as the economic objective and power flow balance equations and N-1 safety checks as operational stability constraints, an operational optimization strategy is generated through dynamic programming. This strategy can reduce operating costs while ensuring the safe and stable operation of the power grid, achieving a balance between economy and stability, and improving the overall efficiency of power grid operation.

[0102] This application addresses the problem that existing scheduling methods cannot effectively handle the impact of grid structure changes on transmission capacity, thus hindering real-time optimization and adjustment of grid operation. It proposes a grid operation optimization method for changes in transmission capacity caused by grid structure variations. This method involves collecting real-time operational data to construct a time-series dataset of the grid structure, extracting topological features to generate a feature matrix, and updating the dynamic topology model. It then quantifies transmission capacity changes using power flow calculations, analyzes the dynamic characteristics of power allocation, extracts key feature vectors of the operating state using principal component analysis, analyzes spatial coordination based on the extracted vectors, generates a spatial optimization representation, and uses a dynamic programming algorithm to generate and execute an operational optimization strategy. This invention can adapt to dynamic changes in the grid structure, enabling real-time optimization and adjustment of the grid operation state and improving grid stability.

[0103] It should be noted that although the steps in the flowchart above are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order requirement for the execution of these steps, and they can be executed in other orders.

[0104] In another embodiment, such as Figure 2 As shown, a second aspect of the present invention provides a power grid operation optimization system for changes in transmission capacity caused by grid structure variations, comprising:

[0105] The grid topology adjustment module 10 is used to acquire real-time operating data of the power grid to construct a grid structure dataset, and to extract topological features from the grid structure dataset using graph theory to obtain a topological feature matrix to update the dynamic topology model of the grid and generate a real-time topology adjustment model.

[0106] The transmission capacity quantification module 20 is used to calculate the power distribution and transmission capacity of each branch in the power grid based on the real-time topology adjustment model and the power flow calculation method, so as to obtain the quantification results of transmission capacity change, analyze the dynamic characteristics of power distribution of each branch, and generate the dynamic characteristics of power distribution of the grid.

[0107] The operation status determination module 30 is used to construct a high-dimensional data processing set by the dynamic characteristics of the power distribution of the grid structure and the quantitative results of the change in the transmission capacity, so as to extract the main feature vectors of the grid structure operation status using the principal component analysis method to obtain the grid structure status feature set.

[0108] The optimization strategy execution module 40 is used to analyze the matching degree between the main feature vector and the branch power distribution based on the grid state feature set, so as to generate a spatial optimization representation of the grid operation state and combine it with the operation stability constraints and economic objectives of the power grid, and use a dynamic programming algorithm to generate an operation optimization strategy to control the power grid execution.

[0109] It should be noted that the modules in the aforementioned power grid operation optimization system for changes in transmission capacity due to grid structure changes can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module. For specific limitations regarding the power grid operation optimization system for changes in transmission capacity due to grid structure changes, please refer to the limitations of the power grid operation optimization method for changes in transmission capacity due to grid structure changes mentioned above; both have the same function and role, and will not be repeated here.

[0110] In summary, this invention relates to the field of information technology and discloses a method and system for optimizing power grid operation when transmission capacity changes due to grid structure variations. It acquires real-time power grid operation data to construct a grid structure dataset, extracts topological features from it to update the dynamic topology model of the grid, calculates transmission capacity changes using power flow calculation methods, and analyzes the dynamic characteristics of power distribution in each branch. Principal component analysis is used to extract key feature vectors characterizing the grid's operating state from the transmission capacity changes and power distribution dynamic characteristics. The matching degree between these feature vectors and branch power distribution is then analyzed to obtain a spatially optimized representation of the grid's operating state. This representation is then combined with power grid operation stability constraints and economic objectives. A dynamic programming algorithm is used to generate an operation optimization strategy to control grid execution. This invention can adapt to dynamic changes in the grid structure, achieve real-time optimization and adjustment of the power grid's operating state, and improve the stability of power grid operation.

[0111] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0112] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this invention, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.

Claims

1. A method for optimizing power grid operation when transmission capacity changes due to grid structure variations, characterized in that, include: Real-time operation data of the power grid is acquired to construct a grid structure dataset, and graph theory is used to extract topological features from the grid structure dataset to obtain a topological feature matrix to update the dynamic topology model of the grid and generate a real-time topology adjustment model. Based on the real-time topology adjustment model, the power distribution and transmission capacity of each branch in the power grid are calculated using the power flow calculation method. The quantitative results of the transmission capacity change are obtained to analyze the dynamic characteristics of power distribution of each branch and generate the dynamic characteristics of power distribution of the grid. A high-dimensional data processing set is constructed by the dynamic characteristics of the power distribution of the grid structure and the quantitative results of the change in the transmission capacity. Principal component analysis is then used to extract the main feature vectors of the grid structure's operating state to obtain the grid structure's state feature set. Based on the network state feature set, the matching degree between the main feature vectors and the branch power distribution is analyzed to generate a spatial optimization representation of the network operation state. This representation is then combined with the operational stability constraints and economic objectives of the power grid. A dynamic programming algorithm is used to generate an operation optimization strategy to control the execution of the power grid. The spatial optimization representation of the generated grid operation state is combined with the grid's operational stability constraints and economic objectives, and a dynamic programming algorithm is used to generate an operation optimization strategy, including: Principal component analysis is used to reduce the dimensionality of the spatial optimization representation to obtain a dimensionality-reduced feature vector. When the dimension of the dimensionality-reduced feature vector is lower than a preset dimension threshold, the dimension of the dimensionality-reduced feature vector is expanded by Gaussian process regression to obtain an expanded feature vector. Based on the extended feature vector, the dynamic response characteristics of the grid structure are determined, and the economic objective is to minimize the operating cost. The operating stability constraints are the power flow balance equation and the N-1 safety check. An operating optimization strategy is generated through a dynamic programming algorithm.

2. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 1, characterized in that, The acquisition of real-time power grid operation data to construct a grid structure dataset includes: Real-time acquisition of node voltage values, branch current values, and topology connection status in the power grid is used to construct an original dataset and preprocess it to obtain a cleaned dataset. The cleaned dataset is constructed into a time series dataset in chronological order, and the dynamic features of the grid structure are extracted using the sliding window method to obtain a grid structure dataset that characterizes the changes in the grid structure.

3. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 1, characterized in that, The method of extracting topological features from the network structure dataset using graph theory to obtain a topological feature matrix includes: Based on the aforementioned network structure dataset, a depth-first search algorithm is used to traverse all nodes in the network to calculate the degree of each node. Ohm's law is used to calculate the weight of each branch, and the Floyd-Warshall algorithm is used to calculate the shortest path between the nodes to generate a connectivity matrix. Based on the degree of each node, the weight of each branch, and the connectivity matrix, an initial topological feature matrix is ​​constructed, and principal component analysis is used to extract the main topological features in the initial topological feature matrix. The K-means clustering algorithm is used to perform cluster analysis on the main topological features to identify nodes and branches with similar topological features, so as to construct node distribution groups. The minimum spanning tree algorithm is then used to optimize the branches in the node distribution groups to obtain optimized branch connection schemes to update the initial topological feature matrix, thus obtaining the final topological feature matrix.

4. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 3, characterized in that, Based on the real-time topology adjustment model, the power distribution and transmission capacity of each branch in the power grid are calculated using the power flow calculation method to obtain the quantitative results of transmission capacity changes, including: Based on the real-time topology adjustment model, and combined with the injected power of each node and the impedance of each branch, the power flow is calculated using the Newton-Raphson method to obtain the power distribution data of each branch. Based on the power distribution data of each branch, the transmission capacity change data of each branch is extracted to generate a transmission capacity quantification dataset. When the transmission capacity of a branch in the transmission capacity quantification dataset is lower than a preset transmission capacity threshold, the weight of each branch is adjusted to generate an optimized branch weight scheme. The real-time topology adjustment model is optimized by the optimized branch weight scheme to obtain an optimized topology adjustment model for re-execution of power flow calculation until the transmission capacity of all branches in the final transmission capacity quantification dataset is not lower than the preset transmission capacity threshold. The final transmission capacity quantification dataset is then output as the transmission capacity change quantification result.

5. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 1, characterized in that, The process of obtaining quantitative results of transmission capacity changes to analyze the dynamic characteristics of power distribution in each branch and generate dynamic characteristics of grid power distribution includes: Based on the quantification results of the transmission capacity change, the power distribution data of each branch is extracted to calculate the correlation between the branch power and the rate of change of transmission capacity, and to generate a dynamic response feature description. If the power change rate of a branch in the dynamic response feature description exceeds a preset power change rate threshold, the dynamic response feature description is adjusted by a weighted average method to obtain an adjusted dynamic response feature description. A power allocation dynamic characteristic model is constructed based on the adjusted dynamic response characteristic description, and the power allocation dynamic characteristics of the grid structure are generated through the power allocation dynamic characteristic model.

6. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 1, characterized in that, The principal component analysis method is used to extract the main feature vectors of the grid structure's operating state, resulting in a grid structure state feature set, including: The high-dimensional data processing set is processed by principal component analysis to extract the main feature vectors characterizing the operation state of the grid structure, and a dynamic characterization of the operation state of the grid structure is generated based on the main feature vectors as the initial grid structure state feature set. If there are major feature vectors in the initial grid state feature set with a variance less than a preset variance threshold, then the initial grid state feature set is updated by linear interpolation to obtain the grid state optimized feature set. Based on the aforementioned network state optimization feature set, a dynamic characterization model of network operation state is constructed to generate a dynamic characterization vector of network topology. Based on the dynamic characterization vector, the dynamic interaction relationship between branch power and network topology is analyzed to obtain a stability feature description characterizing network power allocation. According to the stability feature description, periodic features of the change in transmission capacity are extracted from the high-dimensional data processing set to generate a periodic feature dataset. When there is data in the periodic feature dataset whose change period exceeds a preset change period threshold, the periodic features are decomposed by Fourier transform to obtain a decomposed periodic feature set. The space frame state optimization feature set is updated by decomposing the periodic feature set to obtain the space frame state target feature set as the output space frame state feature set.

7. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 6, characterized in that, After extracting the main feature vectors of the grid structure's operating state using principal component analysis to obtain the grid structure's state feature set, the process also includes: Calculate the variance contribution rate of each of the main feature vectors in the set of state features of the space frame; When it is determined that the variance contribution rate of the main feature vector in the set of network state features is lower than the preset contribution threshold, a deep learning algorithm is used to perform nonlinear dimensionality reduction on the high-dimensional data processing set to obtain a dimensionality-reduced data processing set. The dimensionality-reduced data set is processed using a convolutional neural network to extract features, resulting in the final network structure state feature set for output.

8. The power grid operation optimization method for changes in transmission capacity caused by grid structure changes according to claim 1, characterized in that, The step of analyzing the matching degree between the main feature vectors and the branch power distribution based on the grid state feature set to generate a spatial optimization representation of the grid operating state includes: The spatial distribution characteristics of each main feature vector in the state feature set of the grid structure are calculated using the kernel density estimation method to obtain the feature distribution characterization vector. The probability density function of the power distribution of each branch is then calculated and matched with the feature distribution characterization vector using KL divergence to obtain the matching result. The space where the matching results meet the matching criteria is taken as the target region, and a network topology model is created by using a graph convolutional network. The node voltage and branch impedance of the target region are input into the network topology model for processing, and node embedding vectors are output. The node embedding vectors are spatially partitioned using the DBSCAN clustering algorithm to obtain several highly coordinated regions. The random forest regression method is then used to establish the mapping relationship between the main feature vectors in the target region and each of the highly coordinated regions to generate a spatially optimized representation matrix. The dimensions of the spatial optimization representation matrix are compressed and missing values ​​are filled by bilinear interpolation to generate a spatial coordination heatmap as a spatial optimization representation output of the grid structure's operating status.

9. A power grid operation optimization system for changes in transmission capacity caused by grid structure variations, characterized in that, include: The grid topology adjustment module is used to acquire real-time operating data of the power grid to construct a grid structure dataset, and to extract topological features from the grid structure dataset using graph theory to obtain a topological feature matrix to update the dynamic topology model of the grid and generate a real-time topology adjustment model. The transmission capacity quantification module is used to calculate the power distribution and transmission capacity of each branch in the power grid based on the real-time topology adjustment model and the power flow calculation method, so as to obtain the quantification results of transmission capacity changes, analyze the dynamic characteristics of power distribution of each branch, and generate the dynamic characteristics of power distribution of the grid. The operation status determination module is used to construct a high-dimensional data processing set by using the dynamic characteristics of the power distribution of the grid structure and the quantitative results of the change in the transmission capacity, so as to extract the main feature vectors of the grid structure operation status using the principal component analysis method, and obtain the grid structure status feature set. The optimization strategy execution module is used to analyze the matching degree between the main feature vector and the branch power distribution based on the grid state feature set, so as to generate a spatial optimization representation of the grid operation state and combine it with the operation stability constraints and economic objectives of the power grid. The module uses a dynamic programming algorithm to generate an operation optimization strategy to control the power grid execution. The spatial optimization representation of the generated grid operation state is combined with the grid's operational stability constraints and economic objectives, and a dynamic programming algorithm is used to generate an operation optimization strategy, including: Principal component analysis is used to reduce the dimensionality of the spatial optimization representation to obtain a dimensionality-reduced feature vector. When the dimension of the dimensionality-reduced feature vector is lower than a preset dimension threshold, the dimension of the dimensionality-reduced feature vector is expanded by Gaussian process regression to obtain an expanded feature vector. Based on the extended feature vector, the dynamic response characteristics of the grid structure are determined, and the economic objective is to minimize the operating cost. The operating stability constraints are the power flow balance equation and the N-1 safety check. An operating optimization strategy is generated through a dynamic programming algorithm.

Images