Power Grid Topology Optimization Method Based on Time-Series Data Analysis

Abstract

The present invention discloses a power grid topology optimization method based on time series data analysis, which relates to the technical field of power grid topology optimization and includes the following steps: dividing the power grid into several sub-regions and injecting disturbance signals in the sub-regions; obtaining time series disturbance response data, extracting oscillation characteristics, and identifying oscillation nodes; constructing an oscillation weakness assessment model to conduct weakness assessment on the oscillation nodes; correcting the phase difference matrix of DTW based on the assessment results to screen cross-regional risk propagation paths; and performing power grid topology reconstruction according to the screening results to solve the problems of inaccurate identification of weak nodes and low accuracy of screening cross-regional high-risk paths in traditional network optimization methods.

Description

Technical Field

[0001] The present invention relates to the technical field of power network topology optimization, and more specifically, to a power network topology optimization method based on time series data analysis. Background Art

[0002] With the large-scale grid connection of new energy and the wide access of power electronic devices, modern power systems are facing more complex challenges. The access of new energy such as wind energy and solar energy brings uncertainties and fluctuations in power output, putting greater pressure on the stability of the power grid. At the same time, although the popularization of power electronic devices has improved the flexibility of the power grid, it has also made the dynamic characteristics of the power grid more complex and increased the oscillation risk. Especially under high load and new energy grid connection, the power system is vulnerable to instantaneous current and voltage fluctuations, further exacerbating the complex oscillation problem. In addition, with the increase in grid interconnectivity and inter-regional power exchange, the cross-regional risk propagation paths have become more concealed, and traditional power grid stability analysis methods often cannot effectively address these challenges. Therefore, how to improve the stability and reliability of the power grid, especially the real-time identification and control of complex oscillations and cross-regional risk propagation paths, has become the core problem faced by modern power systems.

[0003] For example, an active distribution network fault recovery method disclosed in the invention patent announcement with the publication number CN117810996B, by modifying the network constraint conditions in the traditional network reconfiguration problem, enables load shedding operations and the generation of islands, and allows island operation and network reconfiguration operations to be coordinated; then, the second-order cone relaxation is used to process the non-linear terms in the model, transforming the original model into a standard mixed-integer second-order cone model; determining the output of the main power supply and distributed power sources to match the load, using an improved Dijkstra algorithm for island division, and then performing network reconfiguration to confirm the load recovery amount; judging whether the distribution network system meets the power flow constraints and outputting the multi-period fault recovery operation results. By reasonably separating the network, avoiding the spread of faults, while reducing the lost load, the reliability of the system is improved by adjusting the power network topology structure and optimizing the energy transmission path.

[0004] For example, the invention patent announcement No. CN118539441B discloses a power grid topology optimization method and system based on search ranking, which includes obtaining power grid topology structure data and constructing an action space based on the power grid topology structure data; wherein the power grid topology structure data includes a node set, an edge set, and an attribute set; obtaining real-time power grid state data, constructing an adaptive two-tower model based on the real-time power grid state data and the action space, and calculating to obtain a candidate action set; constructing a graph attention network and a temporal graph neural network based on the candidate action set to obtain a sorted candidate action list; and performing multi-objective optimization and decision-making based on the sorted candidate action list to obtain an optimal action. Through the innovative algorithm construction and the organic combination of multiple methods, the present invention not only improves the overall effect of power grid topology optimization, but also enhances the adaptability and decision-making quality of the system, realizes the rapid optimization of large-scale power grid systems, and provides strong support for practical applications.

[0005] In the above disclosed technical solution, there are at least the following technical problems:

[0006] In traditional power grid optimization methods, traditional power network optimization methods usually rely on preset power grid topology structures and simplified models. When facing sudden faults, these methods often lack flexible adjustment capabilities. Especially in the screening of cross-regional risk propagation paths, traditional methods cannot respond in real time to changes in the power grid topology structure, resulting in the inability to effectively control the spread of faults. In response to the above problems, the present invention proposes a solution. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a power network topology optimization method based on temporal data analysis. By identifying oscillation nodes and evaluating vulnerability based on sub-region perturbation and temporal data analysis, correcting the phase difference matrix of DTW to screen cross-regional paths, and dynamically reconstructing the power grid topology, the problems of inaccurate identification of weak nodes and low accuracy in screening cross-regional high-risk paths in traditional network optimization methods are solved.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A power network topology optimization method based on temporal data analysis includes the following steps: dividing the power grid into several sub-regions and injecting perturbation signals into the sub-regions; obtaining temporal perturbation response data and extracting oscillation characteristics to identify oscillation nodes; constructing an oscillation weakness evaluation model to evaluate the weakness of the oscillation nodes; correcting the phase difference matrix of DTW based on the evaluation results to screen cross-regional risk propagation paths; and reconstructing the power grid topology according to the screening results.

[0010] In a preferred embodiment, the step of dividing the power grid into several sub-regions and injecting disturbance signals into the sub-regions is specifically as follows: The power grid is divided into several sub-regions, the topological data of the power grid nodes in the sub-regions is obtained, and the PSM matrix of the sub-region power grid nodes is calculated. The PSM matrix is constructed based on the phase difference and time-series response characteristics between the power grid nodes; the power grid nodes are screened based on the PSM matrix of the sub-region power grid nodes to obtain the screened nodes; and disturbance signals are injected into the screened nodes.

[0011] In a preferred embodiment, the step of obtaining time-series disturbance response data, extracting oscillation characteristics, and identifying oscillation nodes is specifically as follows: Time-domain and frequency-domain feature extractions are respectively performed on the time-series disturbance response data to obtain power eigenvalues and communication eigenvalues; the power eigenvalues and communication eigenvalues are fused to obtain the oscillation characteristics of the nodes; and the oscillation characteristics are judged using preset constraint conditions to obtain the oscillation nodes of the nodes.

[0012] In a preferred embodiment, the step of constructing an oscillation weakness evaluation model and performing weakness evaluation on the oscillation nodes is specifically as follows: The derivative of the weakness point energy is calculated and input into a preset physical layer to obtain the physical constraint characteristics of the weakness point; the physical constraint characteristics of the weakness point are input into a preset machine learning model for training; the machine learning model is verified to obtain the oscillation weakness evaluation model; the phase angle time-series data and topological data of each oscillation node are obtained, and the instantaneous kinetic energy component, damping ratio, and maximum phase difference of the oscillation node are calculated; the instantaneous kinetic energy component, damping ratio, and maximum phase difference of the oscillation node are input into the oscillation weakness evaluation model to obtain the weakness value of each oscillation node; and based on the weakness value, weakness evaluation is performed on the oscillation nodes.

[0013] In a preferred embodiment, the step of correcting the phase difference matrix of DTW based on the evaluation result and screening the cross-region risk propagation path is specifically as follows: The phase angle sequence of the nodes within a preset time window is obtained; the cumulative distance between node pairs is calculated and an initial cumulative distance matrix is constructed; the minimum cumulative distance of the phase angle sequences of each pair of node pairs is calculated based on a preset recurrence formula; a first phase difference matrix is constructed based on the minimum cumulative distance; the first phase difference matrix is corrected based on the weakness value of the oscillation node, and the cross-region risk propagation path is screened.

[0014] In a preferred embodiment, the step of correcting the first phase difference matrix based on the weakness value of the oscillation node is specifically as follows: The average value and standard deviation of the weakness values of the oscillation nodes are calculated, and the ratio of the average value and standard deviation of the weakness values is used as the correction weight; it is judged whether the nodes in the node pair are in the same sub-region to obtain a cross-region matrix; and the first phase difference matrix is corrected based on the correction weight and the cross-region matrix to obtain a second phase difference matrix.

[0015] In a preferred embodiment, the screening of the cross-region risk propagation path is specifically as follows: taking the sub-regions as graph nodes, the topological relationship between sub-regions as edges, and the elements of the second phase matrix as edge weights to construct a graph model; using the Dijkstra algorithm to traverse the graph model and screen the cross-region risk propagation path.

[0016] In a preferred embodiment, the use of the Dijkstra algorithm to traverse the graph model and screen the cross-region risk propagation path is specifically as follows: initializing a priority queue, storing paths and path cumulative risk values; based on the cross-region matrix, traversing the nodes in the graph model and updating the priority queue to obtain a list of candidate paths; calculating the first risk value of the candidate paths based on the path cumulative risk values of the candidate paths, and sorting the first risk values of the candidate paths in descending order; comparing and analyzing the first risk value with a preset threshold to update the list of candidate paths and obtain a list of cross-region high-risk paths.

[0017] The technical effects and advantages of the power grid topology optimization method based on time series data analysis of the present invention:

[0018] 1. By dividing the power grid into several sub-regions and injecting perturbation signals within the sub-regions, the present invention can simulate various possible network states and comprehensively capture the dynamic responses of different parts of the power grid. This refined local perturbation method not only helps to improve the accuracy of data, but also can discover potential oscillation nodes and weak links by screening key nodes, providing accurate data support for subsequent optimization.

[0019] Secondly, in the method, by combining time-domain and frequency-domain feature extraction to obtain power and communication features and performing feature fusion, the oscillation features of nodes are effectively extracted. By presetting constraint conditions, oscillation nodes are quickly identified, laying a foundation for further weak link assessment. The oscillation weak link assessment model provides a more accurate assessment of weak nodes through the combination of physical layer constraint features and machine learning, can dynamically evaluate the stability of each node in the power grid, and timely identify potential risk points.

[0020] In addition, based on the correction of the DTW (Dynamic Time Warping) phase difference matrix and the screening of the cross-region risk propagation path, the response speed and accuracy of the system are further improved. By correcting the phase difference matrix and combining the weak values of oscillation nodes, the cross-region risk propagation path can be dynamically adjusted, effectively avoiding the spread of potential risks in the network and ensuring the stability of the system. In particular, using the Dijkstra algorithm to screen cross-region high-risk paths can identify potential high-risk paths in real time in the changing power grid topology, helping the power grid to adjust and optimize in a timely manner when facing emergencies, thereby improving the reliability of the power grid. Brief Description of the Drawings

[0021] Figure 1 This is a schematic flowchart of the power network topology optimization method based on time series data analysis of the present invention. Specific embodiments

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] Embodiment 1 Figure 1 The power network topology optimization method based on time series data analysis of the present invention is given, including the following steps:

[0024] S1, divide the power grid into several sub-regions and inject disturbance signals into the sub-regions;

[0025] In this example, dividing the power grid into several sub-regions and injecting disturbance signals into the sub-regions specifically are:

[0026] Divide the power grid into several sub-regions, obtain the topology data of the power grid nodes in the sub-regions, and calculate the PSM matrix of the sub-region power grid nodes. The PSM matrix is constructed based on the phase difference and time series response characteristics between the power grid nodes;

[0027] Screen the power grid nodes based on the PSM matrix of the sub-region power grid nodes to obtain the screened nodes;

[0028] Inject disturbance signals into the screened nodes.

[0029] It should be noted that dividing the power grid into several sub-regions and injecting disturbance signals into the sub-regions, adopting a refined regional management strategy, aims to improve the stability and response ability of the power system. The specific implementation steps include: first, divide the power grid into multiple sub-regions, which are relatively independent in topological structure and help to analyze the dynamic response of the power grid locally. In each sub-region, obtain the topology data of the power grid nodes in this region and calculate the corresponding PSM matrix. This process is a quantification and characterization of the power grid structure and the mutual relationship between nodes.

[0030] The PSM matrix is constructed based on the phase differences and time - series response characteristics between power grid nodes, reflecting the phase change characteristics of each node in the power grid. By calculating the PSM matrix, the coupling relationships and dynamic behaviors between various nodes in the power grid can be accurately identified. Then, based on this PSM matrix, the power grid nodes are screened to identify the key nodes that have a greater impact on the power grid stability. This screening process is of great significance because it can help determine the potential weak links or high - risk areas in the power grid, providing a basis for subsequent optimization and adjustment.

[0031] Next, a perturbation signal is injected into the screened key nodes. The core of this step is to simulate the dynamic response of the power grid under perturbations, which can reflect the stability and anti - interference ability of each node when facing external perturbations. By injecting the perturbation signal and monitoring the system response, a large amount of time - series data can be obtained, providing data support for subsequent oscillation feature extraction, weak link assessment, and topology optimization. The advantage of this method is that by performing local perturbations within the sub - regions, it can effectively avoid the excessive impact of large - scale perturbations on the entire power grid and can more precisely analyze the stability of each node in the power grid.

[0032] In addition, the method of dividing the power grid into sub - regions and injecting local perturbations avoids the global perturbation of the entire power grid in traditional methods, reduces the risks brought by perturbations, and makes the analysis results more specific and operable. In this way, power grid optimization does not solely rely on the analysis of global data but can conduct dynamic tests and analyses in multiple local regions, thereby discovering potential weak links and carrying out targeted optimizations.

[0033] S2. Obtain time - series perturbation response data, extract oscillation features, and identify oscillation nodes.

[0034] In this example, obtaining time - series perturbation response data, extracting oscillation features, and identifying oscillation nodes are specifically as follows:

[0035] Extract time - domain and frequency - domain features from the time - series perturbation response data respectively to obtain power eigenvalues and communication eigenvalues.

[0036] Fuse the power eigenvalues and communication eigenvalues to obtain the oscillation features of the nodes.

[0037] Judge the oscillation features using preset constraint conditions to obtain the oscillation nodes of the nodes.

[0038] It should be noted that time - series disturbance response data is subjected to time - domain and frequency - domain feature extraction to obtain the dynamic characteristics of the power system from two levels of time and frequency respectively. In time - domain analysis, the dynamic response characteristics, fluctuation amplitude, and waveform changes of the system are used to reflect the stability of grid nodes; while in frequency - domain analysis, information such as the frequency characteristics and harmonic components of the system are mainly concerned to help identify possible resonance phenomena or frequency fluctuations in the power grid. Through these two feature - extraction methods, the oscillation characteristics of different nodes in the power system can be comprehensively captured, providing high - quality input data for subsequent analysis.

[0039] Then, the power eigenvalue and the communication eigenvalue are fused to obtain a comprehensive node oscillation feature. The power eigenvalue mainly reflects the electrical characteristics of grid nodes, such as voltage, power, etc., while the communication eigenvalue focuses on information at the network level such as communication delay and transmission loss between nodes. By fusing these two types of features, the response behavior of grid nodes under disturbances can be more comprehensively characterized. Especially in modern power systems, the stability of communication and information transmission plays a crucial role in the overall stability of the power grid. After feature fusion, the oscillation characteristics are more accurate, providing stronger support for the subsequent identification of oscillation nodes.

[0040] After obtaining the oscillation features, the method judges the oscillation features through preset constraint conditions to identify the oscillation nodes in the power grid. This step determines which nodes are oscillation nodes based on the oscillation features and response characteristics of the nodes, combined with the physical constraints of the system, such as voltage limits and power fluctuation ranges. These oscillation nodes are often the weak links in the power grid that are most vulnerable to disturbances or prone to causing faults. By timely identifying and carrying out targeted interventions, large - scale faults can be effectively avoided.

[0041] S3. Construct an oscillation weakness assessment model to assess the weakness of oscillation nodes;

[0042] In this example, constructing an oscillation weakness assessment model to assess the weakness of oscillation nodes specifically includes:

[0043] Calculate the derivative of the weakness - point energy and input it into the preset physical layer to obtain the physical - constraint features of the weakness point;

[0044] Input the physical - constraint features of the weakness point into the preset machine - learning model for training;

[0045] Verify the machine - learning model to obtain the oscillation weakness assessment model;

[0046] Obtain the phase - angle time - series data and topological data of each oscillation node and calculate the instantaneous kinetic - energy component, damping ratio, and maximum phase difference of the oscillation node;

[0047] Input the instantaneous kinetic energy component, damping ratio, and maximum phase difference of the oscillation nodes into the oscillation vulnerability assessment model to obtain the vulnerability value of each oscillation node;

[0048] Based on the vulnerability values, conduct a vulnerability assessment of the oscillation nodes.

[0049] It should be noted that by calculating the derivative of the vulnerability point energy and inputting it into the preset physical layer, the physical constraint characteristics of the vulnerability point are obtained. This step combines the physical laws of the power grid, such as current, voltage, etc., and can accurately reflect the energy change and its physical characteristics of each node under disturbances. This process fully reveals the physical characteristics of the oscillation nodes and helps to analyze which nodes show greater instability or susceptibility under external disturbances or load changes.

[0050] Next, these physical constraint characteristics are input into the preset machine learning model for training. Through the learning of the machine learning model, the deep - level relationship between node oscillation and vulnerability can be discovered from historical data. The machine learning method has the characteristics of self - adaptation and high efficiency, and can automatically adjust the assessment model according to the actual operation of the power grid, thereby improving the accuracy and robustness of the vulnerability assessment. After training, a reliable oscillation vulnerability assessment model is obtained through model verification, providing a theoretical basis for further assessment of vulnerable nodes.

[0051] Next, the method calculates the instantaneous kinetic energy component, damping ratio, and maximum phase difference of each vulnerable node by obtaining the phase - angle time - series data and topological data of each oscillation node. These dynamic characteristics can reveal the kinetic energy change, oscillation intensity, and phase difference of the power - grid nodes under disturbances, thereby further quantifying the stability of the nodes. Based on these characteristics, they are input into the oscillation vulnerability assessment model for prediction to obtain the vulnerability value of each node. The larger the vulnerability value, the more vulnerable the node is to disturbances in the power grid and the more likely it is to cause abnormal operation of the system.

[0052] By conducting a vulnerability assessment of the oscillation nodes based on the vulnerability values, the vulnerable links in the power grid can be accurately identified, and targeted measures can be taken in a timely manner for optimization and adjustment. This process can not only improve the stability of the power grid but also early - warning potential fault risks, thereby reducing the occurrence of large - scale fault events and optimizing the safety and reliability of power - grid operation.

[0053] To sum up, this method based on the oscillation vulnerability assessment model effectively realizes the dynamic assessment of the vulnerability of power - grid nodes by combining the feature analysis at the physical level and the intelligent processing of the machine learning model. Through this precise assessment method, the power system can achieve more efficient risk prediction and early - warning, greatly improving the reliability and anti - interference ability of the power grid, and providing a solid technical support for the intelligent management and optimal dispatching of modern power grids.

[0054] Among them, the instantaneous kinetic energy component, damping ratio, and maximum phase difference of the oscillation nodes are input into the oscillation weakness assessment model to obtain the weakness value of each oscillation node. The specific calculation formula is as follows:

[0055]

[0056] Among them, is the weakness value of oscillation node i at time t, is the instantaneous kinetic energy component of oscillation node i, is the preset reference kinetic energy, is the damping ratio of oscillation node i, is the preset critical damping ratio threshold, is the maximum phase difference of oscillation node i.

[0057] S4. Based on the evaluation results, correct the phase difference matrix of DTW and screen the cross-regional risk propagation paths;

[0058] In this example, based on the evaluation results, correct the phase difference matrix of DTW and screen the cross-regional risk propagation paths. Specifically:

[0059] Obtain the phase angle sequence of the nodes within the preset time window;

[0060] Calculate the cumulative distance between node pairs and construct an initial cumulative distance matrix;

[0061] Based on the preset recurrence formula, calculate the minimum cumulative distance of the phase angle sequences of each pair of node pairs;

[0062] Construct the first phase difference matrix based on the minimum cumulative distance;

[0063] Based on the weakness value of the oscillation nodes, correct the first phase difference matrix and screen the cross-regional risk propagation paths.

[0064] It should be noted that by correcting the dynamic time warping (DTW) phase difference matrix based on the oscillation weakness assessment results and accurately screening the cross-regional risk propagation paths, the stability analysis and risk prediction capabilities of the power system under complex disturbance conditions can be effectively improved. Specifically, the core of this method is to analyze the phase difference between nodes based on the phase angle sequences of power grid nodes and combine the DTW algorithm to reveal the potential risk propagation paths between different regions in the power grid. The following are the detailed steps and advantages of this process.

[0065] First, the method obtains the phase angle sequences of nodes within a preset time window. The phase angle sequences reflect the relative voltage changes of each node in the power grid, which are important data for evaluating the dynamic response and stability of the power grid. This step can accurately track the response process of the power grid during a disturbance by capturing the dynamic changes of nodes in the time series.

[0066] Next, the method calculates the cumulative distances between node pairs and constructs an initial cumulative distance matrix. This process quantifies the degree of dynamic correlation between nodes by calculating the distances between phase angle sequences. The initial cumulative distance matrix provides a basis for subsequent phase difference calculations. It can show the degree of phase angle fluctuations between nodes and lays a data foundation for further risk analysis.

[0067] Then, based on a preset recurrence formula, the method calculates the minimum cumulative distance of the phase angle sequences between each pair of nodes. This process is the core step of the DTW algorithm. By using the recurrence formula to find the minimum cumulative distance path, it ensures that the phase difference matrix can accurately reflect the phase change relationship between nodes. The minimum cumulative distance is an important basis for evaluating the risk propagation paths between nodes in the power grid because it can quantify the degree of linkage and disturbance transfer efficiency between nodes.

[0068] On this basis, the method constructs a first phase difference matrix based on the minimum cumulative distance, further revealing the phase relationship between nodes and its potential risk propagation paths. The construction of the first phase difference matrix provides a reference basis for subsequent correction and optimization, enabling the spatio-temporal correlation characteristics of each node in the power grid to be clearly presented.

[0069] Next, the method corrects the first phase difference matrix according to the weakness value of the oscillating nodes. The key to this process is that the weakness value can reflect the stability of the oscillating nodes. By correcting the phase difference matrix, it can dynamically adjust the phase relationship in the high-risk areas of the power grid, making the screening of risk propagation paths more accurate. Through this correction, the risk of misjudgment and missed judgment can be reduced, providing more accurate path data for subsequent screening of cross-regional risk propagation paths.

[0070] Finally, based on the corrected phase difference matrix, the method screens out the cross-regional risk propagation paths in the power grid. Through this screening process, potential risk transfer paths in the power grid can be identified, especially the linkage effects between different regions, helping decision-makers to timely identify cross-regional risk propagation channels and take effective preventive measures. The screened high-risk paths will become the focus of power grid optimal dispatching and emergency management to avoid large-scale system failures and accidents.

[0071] The main advantages of this method are as follows: First, it combines the dynamic matching ability of the DTW algorithm, which can effectively analyze the complex time-series relationships between power grid nodes and improve the accuracy of identifying cross-regional risk propagation paths. Second, the correction process of weak nodes makes the phase difference matrix more conform to the actual operating state of the power grid, ensuring more accurate screening of risk propagation paths. Finally, this method can not only identify potential weak nodes in the power grid but also deeply analyze the cross-regional dynamic response of the power grid, thus providing strong data support for the dynamic dispatching and emergency management of the power grid.

[0072] In this example, the first phase difference matrix is corrected based on the weak values of the oscillating nodes, specifically as follows:

[0073] Calculate the average value and standard deviation of the weak values of the oscillating nodes, and use the ratio of the average value and standard deviation of the weak values as the correction weight;

[0074] Judge whether the nodes in the node pair are in the same sub-region to obtain the cross-regional matrix;

[0075] Correct the first phase difference matrix based on the correction weight and the cross-regional matrix to obtain the second phase difference matrix.

[0076] Among them, correcting the first phase difference matrix based on the correction weight and the cross-regional matrix to obtain the second phase difference matrix, the specific calculation formula is as follows:

[0077]

[0078] Among them, is the second phase difference matrix, is the first phase difference matrix, is the correction weight, is the cross-regional matrix of the node pair (m, n), is the electrical distance between node m and node n.

[0079] It should be noted that first, the method calculates the average value and standard deviation of the weak values of the oscillating nodes. The weak values of the oscillating nodes reflect the stability and anti-interference ability of the nodes in the power grid. The higher the weak value, the more likely the node is to be disturbed or cause system failures. Calculating the average value and standard deviation of the weak values of the oscillating nodes helps to quantify the risk levels of the oscillating nodes in the power grid and provides a basis for further correcting the phase difference matrix. By calculating the standard deviation of the weak values, the degree of difference in stability between different nodes can be evaluated, thus assigning different correction weights to each node.

[0080] Next, the method uses the ratio of the average value to the standard deviation of the weak values as the correction weight. This ratio can reflect the degree of risk contribution of each node in the power grid. Nodes with larger weak values and higher standard deviations will be assigned higher correction weights. This step can dynamically adjust the impact of weak nodes on the overall stability of the power grid, ensuring that the risks of weak nodes are appropriately amplified while the influence of stable nodes is suppressed when correcting the phase difference matrix, thus achieving a more accurate risk assessment.

[0081] Then, the method obtains the cross-region matrix by determining whether the nodes in the node pair are in the same sub-region. The power system is often composed of multiple regions, and there may be large dynamic differences between nodes in different regions. By analyzing whether the nodes are in the same sub-region, it can help clarify the coupling relationship between different regions. Especially in the analysis of cross-region risk propagation, determining whether it is cross-region propagation becomes a crucial factor. The construction of the cross-region matrix can effectively distinguish the risk propagation modes within and between regions, thus providing clear regional division information for subsequent risk assessment.

[0082] Based on the correction weight and the cross-region matrix, the method further corrects the first phase difference matrix to obtain the second phase difference matrix. By combining the weak value correction weight and the cross-region matrix and applying them to the correction of the phase difference matrix, the phase difference relationship between nodes in the power grid can be dynamically adjusted. Especially between nodes with cross-region connections, potential risk propagation channels can be accurately captured. This correction process effectively avoids the overly balanced phase difference calculation in traditional methods and can more precisely identify those cross-region propagation paths with higher risks and prone to large-scale failures.

[0083] The main advantages of this method are as follows: First, by calculating the ratio of the weak value and the standard deviation of the oscillating nodes as the correction weight, it can more accurately reflect the risk level of each node and dynamically adjust the stability influence between nodes during the correction of the phase difference matrix. Second, the introduction of the cross-region matrix helps to clearly distinguish the mutual influence between different regions in the power grid and avoid misjudgment of large-scale cross-region propagation paths. In addition, the phase difference matrix based on weak value correction can more accurately screen out high-risk nodes and cross-region risk propagation paths in the power grid, improving the anti-interference ability and fault warning ability of the power grid.

[0084] In this example, the cross-region risk propagation paths are screened as follows:

[0085] Taking the sub-region as the graph node, the topological relationship between sub-regions as the edge, and the elements of the second phase matrix as the edge weight, a graph model is constructed;

[0086] Using the Dijkstra algorithm to traverse the graph model to screen the cross-region risk propagation paths.

[0087] It should be noted that first, the power grid is divided into multiple sub - regions, each sub - region is regarded as a node in the graph, and the topological relationship between sub - regions constitutes the edges of the graph. The elements of the second - phase difference matrix are used as the weights of the edges, reflecting the risk propagation intensity between different sub - regions. The construction of this graph model can effectively represent the regionalized structure of the power grid and the dynamic coupling relationship between its nodes.

[0088] Next, the Dijkstra algorithm is used to traverse the constructed graph model to screen out the cross - region risk propagation paths. The Dijkstra algorithm is a classic algorithm widely used in graph optimization, which can effectively identify the optimal propagation path between nodes by calculating the shortest path. In this process, the weight of the edge (i.e., the value in the phase - difference matrix) determines the risk level of the path, and the algorithm will automatically identify the high - risk cross - region paths. This step can accurately find the potential high - risk propagation channels in the power grid to ensure that possible risk diffusion can be detected and addressed in a timely manner during the operation of the power grid.

[0089] In this example, the Dijkstra algorithm is used to traverse the graph model and screen the cross - region risk propagation paths, specifically as follows:

[0090] Initialize the priority queue, store paths, and path cumulative risk values;

[0091] Based on the cross - region matrix, traverse the nodes in the graph model and update the priority queue to obtain a list of candidate paths;

[0092] Based on the path cumulative risk values of the candidate paths, calculate the first risk values of the candidate paths and sort the first risk values of the candidate paths in descending order;

[0093] Compare and analyze the first risk values with a preset threshold, update the list of candidate paths, and obtain a list of cross - region high - risk paths.

[0094] It should be noted that by initializing the priority queue, storing paths, and path cumulative risk values, preparations are made for subsequent path screening. The priority queue is used to efficiently manage and update candidate paths so that the optimal path selection can be obtained in a timely manner during the traversal process. The path cumulative risk value is used to quantify the cumulative risk degree on each path, and this value is directly related to the operating safety of the power grid. By introducing the priority queue, the path search process can be optimized and the efficiency of path screening can be improved.

[0095] Next, based on the cross-region matrix, the nodes in the graph model are traversed, and the priority queue is updated to obtain a list of candidate paths. The cross-region matrix is used to represent the risk coupling relationship between different sub-regions, and the traversal of nodes and the update of the priority queue ensure that the linkage risks between different regions can be fully reflected. In this way, the risk propagation channels between different regions of the power grid can be systematically analyzed, providing a wide range of candidate path information for subsequent path screening.

[0096] Then, based on the path cumulative risk value of the candidate paths, the first risk value of each candidate path is calculated, and the first risk values of the candidate paths are sorted in descending order. The first risk value of a path is calculated based on the cumulative risk value, reflecting the total risk level of each path. By sorting the path risk values, it is possible to clearly identify which paths have higher risks, and then focus on those high-risk propagation channels that may trigger large-scale power grid failures. The descending order sorting ensures that higher-risk paths are given priority attention, providing an important reference for subsequent decision-making.

[0097] Finally, the calculated first risk values are compared and analyzed with a preset threshold, and the candidate path list is updated based on the threshold setting to obtain a list of cross-region high-risk paths. The setting of this threshold is adjusted according to the fault tolerance ability, stability requirements, and historical fault data of the power grid, ensuring that the identification of high-risk paths is more in line with the actual needs of the power grid. Through this process, the finally selected cross-region high-risk paths will contribute to the optimal dispatching and fault emergency response of the power grid.

[0098] The main advantage is that it can accurately identify high-risk propagation paths in the power grid by combining the Dijkstra algorithm with the cumulative risk value and threshold analysis of paths. This not only improves the accuracy of power grid risk assessment but also effectively enhances the real-time and predictive nature of risk management. Through the dynamic update of the priority queue and path risk values, the method can quickly respond to changes in power grid operation, ensuring that cross-region risk propagation paths can be quickly and accurately screened in a complex power grid environment. This process not only helps to improve the safety and stability of the power grid but also provides a scientific basis for power grid fault prevention and emergency management, enabling effective measures to be taken quickly when emergencies occur to reduce the scope and impact of risk diffusion.

[0099] S5. Perform power grid topology reconstruction according to the screening results.

[0100] In this example, power grid topology reconstruction is performed according to the screening results, specifically as follows:

[0101] According to the selected cross - regional high - risk propagation paths, the process of power grid topology reconstruction mainly involves identifying high - risk areas and weak nodes, and adjusting the connection mode of the power grid to improve the stability and risk - resistance ability of the power grid. First, determine the high - risk areas in the power grid, especially the weak nodes. These nodes may become key points for fault propagation due to excessive load, equipment aging, or unreasonable structure. Then, take measures to optimize the power grid connection mode. For example, add redundant paths to ensure that when a node fails, other paths can take over the load and avoid large - scale power outages; temporarily disconnect high - risk paths to prevent the spread of faults; and through reasonable load redistribution, relieve the pressure on weak nodes. Power grid topology reconstruction can also be dynamically adjusted in combination with a real - time monitoring system. During the operation of the power grid, the topology structure can be optimized in a timely manner based on real - time risk assessment data, thereby improving the fault - tolerance ability and emergency response speed of the power grid. In addition, by optimizing the power grid topology structure, the operation efficiency of the power grid can be improved, avoiding over - reliance on a certain part of the system, and enhancing the stability and reliability of the power grid. Finally, power grid topology reconstruction not only improves the anti - disturbance ability of the power grid, reduces the potential risk of fault propagation, but also enhances the emergency response ability of the power grid to emergencies, ensuring the efficient and safe operation of the power system.

[0102] The above - mentioned formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0103] The above - mentioned embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above - mentioned embodiments can be implemented in whole or in part in the form of a computer program product.

[0104] Those of ordinary skill in the art can realize that the modules and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0105] In addition, in each embodiment of this application, the functional modules can be integrated into a processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0106] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

[0107] Finally: The above description is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A power network topology optimization method based on time series data analysis, characterized in that: The following steps are involved: The power grid is divided into several sub-areas, and disturbance signals are injected into the sub-areas; Obtain timing disturbance response data and extract oscillation features to identify oscillation nodes; Calculate the derivative of the weak point energy and input it into the preset physical layer to obtain the physical constraint characteristics of the weak point; Input the physical constraint characteristics of the weak points into the preset machine learning model for training and verification to obtain the shock weakness assessment model; Obtain the phase angle timing data and topological data of each oscillation node and calculate the instantaneous kinetic energy component, damping ratio and maximum phase difference of the oscillation node; The instantaneous kinetic energy component, damping ratio and maximum phase difference of the oscillation node are input into the oscillation weakness assessment model to obtain the weakness value of each oscillation node; Based on the weakness value, the oscillation node is evaluated for weakness; Based on the evaluation results, the DTW phase difference matrix is ​​modified to screen the cross-regional risk transmission path; The grid topology is reconstructed according to the screening results.

2. The power network topology optimization method based on time series data analysis according to claim 1 is characterized in that: The power grid is divided into several sub-areas, and disturbance signals are injected into the sub-areas, specifically: Divide the power grid into several sub-areas, obtain topological data of the power grid nodes in the sub-areas, and calculate the PSM matrix of the power grid nodes in the sub-areas, wherein the PSM matrix is ​​constructed based on the phase difference and timing response characteristics between the power grid nodes; Screening the power grid nodes based on the PSM matrix of the sub-region power grid nodes to obtain the screened nodes; Inject disturbance signals into the screened nodes.

3. The method for optimizing power network topology based on time series data analysis according to claim 2, characterized in that: The acquisition of the timing disturbance response data and the extraction of the oscillation characteristics to identify the oscillation nodes is specifically as follows: Extract time domain and frequency domain features of the time series disturbance response data to obtain power characteristic values ​​and communication characteristic values; The power characteristic value and the communication characteristic value are integrated to obtain the oscillation characteristics of the node; The oscillation characteristics are judged using preset constraints to obtain the oscillation node of the node.

4. The method for optimizing power network topology based on time series data analysis according to claim 3 is characterized in that: The DTW phase difference matrix is ​​modified based on the evaluation results to screen the cross-regional risk transmission path, specifically: Obtain the phase angle sequence of the node within the preset time window; Calculate the cumulative distance of node pairs and construct the initial cumulative distance matrix; The minimum cumulative distance of the phase angle sequence of each pair of nodes is calculated based on a preset recursive formula; Constructing a first phase difference matrix based on the minimum cumulative distance; The first phase difference matrix is ​​corrected based on the weak value of the oscillation node, and the cross-regional risk propagation path is screened.

5. The method for optimizing power network topology based on time series data analysis according to claim 4 is characterized in that: The first phase difference matrix is ​​corrected based on the weak value of the oscillation node, specifically: Calculate the mean and standard deviation of the weak values ​​of the oscillating nodes, and use the ratio of the mean and standard deviation of the weak values ​​as the correction weight; Determine whether the nodes in the node pair are in the same sub-region and obtain the cross-region matrix; The first phase difference matrix is ​​corrected based on the correction weight and the cross-region matrix to obtain a second phase difference matrix.

6. The method for optimizing power network topology based on time series data analysis according to claim 5, characterized in that: The screening of cross-regional risk transmission paths is specifically as follows: The graph model is constructed by taking the sub-regions as graph nodes, the topological relationships between the sub-regions as edges, and the elements of the second phase matrix as edge weights; The Dijkstra algorithm is used to traverse the graph model and screen the cross-regional risk transmission paths.

7. The method for optimizing power network topology based on time series data analysis according to claim 6, characterized in that: The Dijkstra algorithm is used to traverse the graph model and screen the cross-regional risk transmission path, specifically: Initialize the priority queue, store the path and the path cumulative risk value; Based on the cross-region matrix, the nodes in the graph model are traversed and the priority queue is updated to obtain a list of candidate paths; Based on the path cumulative risk values ​​of the candidate paths, first risk values ​​of the candidate paths are calculated, and the first risk values ​​of the candidate paths are sorted in descending order; The first risk value is compared and analyzed with a preset threshold, and the candidate path list is updated to obtain a cross-regional high-risk path list.

8. The method for optimizing power network topology based on time series data analysis according to claim 7, characterized in that: The instantaneous kinetic energy component, damping ratio and maximum phase difference of the oscillation node are input into the oscillation weakness assessment model to obtain the weakness value of each oscillation node. The specific calculation formula is as follows: in, is the weak value of the oscillating node i at time t, is the instantaneous kinetic energy component of the oscillation node i, is the preset reference kinetic energy, is the damping ratio of the oscillation node i, is the preset critical damping ratio threshold, is the maximum phase difference of the oscillation node i.

9. The method for optimizing power network topology based on time series data analysis according to claim 8, characterized in that: The first phase difference matrix is ​​corrected based on the correction weight and the cross-region matrix to obtain the second phase difference matrix. The specific calculation formula is as follows: in, is the second phase difference matrix, is the first phase difference matrix, To correct the weight, is the cross-region matrix of node pairs (m,n), is the electrical distance between node m and node n.

Images