Multi-stage decision recursive optimization method for topology identification of high-confidence station area

Abstract

The technical scheme of the invention discloses a multi-stage decision recursive optimization method for topology identification of a high-confidence station area, which is used for solving the problem of unknown key topology in state perception and analysis of a power system. According to the topology identification method based on the energy conservation principle provided by the invention, the phase sequence attribution of the user can be accurately identified only based on the existing low-frequency measurement data by analyzing the electric quantity balance relationship between the user and the transformer, and the method has the remarkable advantages that the data is easy to obtain, additional investment is not needed, and the deployment cost is low. In order to further solve accidental misjudgment caused by small difference of electricity consumption behaviors of users, a cross-period consistency verification strategy is introduced. According to the strategy, the idea of ensemble learning is used for reference, the recognition results of a plurality of time sections are comprehensively evaluated, random disturbance is eliminated, a stable relation is reserved, full-automatic topology recognition with the precision ratio close to 100% is finally achieved, and manual rechecking is not needed.

Description

Technical Field

[0001] This invention relates to a multi-stage decision recursive optimization method based on energy conservation constraints and cross-time consistency analysis for high-confidence power station topology identification, which is used to solve the key topological unknown problem in power system state perception and analysis, and belongs to the field of power system state perception and intelligent analysis. Background Technology

[0002] Currently, transformer substation topology information mainly relies on manual surveying or user-reported data, which generally suffers from three major bottlenecks: high cost, slow updates, and low accuracy, making it difficult to meet the needs of modern distribution networks for accurate and real-time sensing of operational status. Although some studies have attempted to use the similarity of physical quantities such as voltage and current for topology identification, these methods rely on high-frequency synchronous acquisition equipment, limiting their large-scale application in existing low-frequency metering systems. Summary of the Invention

[0003] The purpose of this invention is to provide a transformer area topology identification method based on energy conservation and time consistency analysis. This method utilizes existing low-frequency metering data and dynamic programming algorithms to achieve automatic identification of user phase sequence without the need for additional hardware, and combines ensemble learning to improve accuracy and stability.

[0004] To achieve the above objectives, the technical solution of this invention discloses a multi-stage decision-making recursive optimization method for high-confidence power station topology identification, used to solve the key topology unknown problem in power system state perception and analysis. The method is characterized by the following steps: Let M be different transformer substations (M=3) representing the specific phases of the three-phase power supply connected to the user. Each substation contains K users. Let matrix X represent the electricity consumption of the K users in the current substation over L time periods. Then:

[0005] in: This represents the electricity consumption of K users within the current distribution area during the l-th time period. ; This describes the electricity consumption of the k-th user during the L-th time period. ; Let matrix Y represent the phase sequence attribution of K users, then we have:

[0006] in, ; Let matrix W represent the phase-specific electrical quantities of the transformer, then we have:

[0007] in, Let K be the electricity consumption of the k-th user in the m-th transformer area. ; Line loss matrix Represented as:

[0008] in: Let m be the line loss rate of the m-th transformer station for the k-th user. This represents the total electricity consumption of the k-th user in the m-th transformer substation. Based on the law of conservation of energy, the following relationship is established:

[0009] Treating user phase affiliation Y as the variable to be optimized, we use dynamic programming to solve the above equation, and search for all possible combinations of user phases using an algorithm. The goal is to minimize the following loss function:

[0010] After obtaining the electricity consumption information of K users over L time periods and the phase-specific electricity consumption of each transformer, a dynamic programming algorithm is used to solve the phase sequence attribution relationship of the transformer area.

[0011] Preferably, the total electricity consumption of the k-th user in the m-th transformer substation is... .

[0012] Preferably, in the dynamic programming algorithm, the l-th time interval of length T is divided into q time subsets. The topological assignment results for each time subset are calculated separately. The intersection of the results of one subset is taken as the high confidence judgment, and the other structures not in the intersection are assigned probabilities based on their frequency of occurrence in each subset result.

[0013] Preferably, in the dynamic programming algorithm: if the current user belongs to the same station area in all time subsets, then it is determined that the current user belongs to the phase corresponding to that station area 100%; otherwise, the current user is determined by the frequency of the current user's affiliation in each time subset: if the frequency of the current user's affiliation to a certain station area in all time subsets exceeds the threshold, then the current user belongs to the phase corresponding to the current station area; other cases are considered low confidence and require manual verification.

[0014] Preferably, an overlapping sliding window is used to sample the l-th time interval of length T to obtain q time subsets. .

[0015] This invention proposes a topology identification method based on the principle of energy conservation. By analyzing the power balance relationship between users and transformers, it can accurately identify the phase sequence affiliation of users using only existing low-frequency metering data. This method boasts significant advantages such as readily available data, no additional investment required, and low deployment costs. To further address occasional misjudgments caused by small differences in user electricity consumption behavior, this invention introduces a cross-time-period consistency verification strategy. This strategy draws on the idea of ​​ensemble learning, comprehensively evaluating the identification results across multiple time segments, eliminating random disturbances, and preserving stable relationships. Ultimately, it achieves nearly 100% precision in fully automated topology identification without manual verification. The method disclosed in this invention constructs the topology structure starting from data elements, possessing high accuracy, high stability, and high economy, making it particularly suitable for the large-scale deployment and intelligent evolution needs of distribution networks. Attached Figure Description

[0016] Figure 1 This illustrates the three-layer structure of a low-voltage distribution network; Figure 2 The process of solving time consistency is illustrated. Detailed Implementation

[0017] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0018] Figure 1 The topology of the low-voltage distribution network is shown. When the topology is unknown, the data-based topology identification of the transformer area needs to solve the following problems: (1) Determine the user relationship within the transformer area, that is, the identification of the "transformer-user relationship" of the transformer area; (2) Determine the specific phase (A, B, C) of the three-phase power connected to each user, that is, the identification of the "phase sequence relationship" of the transformer area.

[0019] Based on the premise of energy conservation, the identification of topological relationships in transformer substations is essentially a multi-subset partitioning problem.

[0020] First, for identifying the "change of user relationship" within a transformer substation, the task is to divide the users into subsets belonging to the same substation. For a line U with N users, the electricity consumption of each user on the line is:

[0021] In the formula, Let represent the electricity consumption of the nth user.

[0022] It has M substations under its jurisdiction, and the total electricity consumption of each substation is:

[0023] In the formula, Let m be the electricity consumption of the m-th transformer area.

[0024] Then, we need to divide these N users into M mutually exclusive non-empty subsets. , make each subset The internal energy satisfies the law of conservation of electric charge, that is:

[0025] when At this point, the problem degenerates into the "phase sequence recognition" problem of the transformer substation, in which... , , These represent the total electricity consumption of phases A, B, and C, respectively. The goal is to divide all users into three mutually exclusive non-empty subsets. , , This ensures that the total charge of each subset is as close as possible to the total charge of its corresponding phase.

[0026] Taking the user phase sequence identification problem as an example, assuming there are K users in a transformer area, the electricity consumption of these users can be represented by matrix X over L time periods:

[0027] in: This shows the electricity consumption of K users within the current distribution area during the l-th time period (usually the l-th day). ; This describes the electricity consumption of the k-th user during the L-th time period. .

[0028] At the same time, a matrix Y can be constructed to represent the phase sequence affiliation of K users:

[0029] Therefore, matrix Y should satisfy: Then the total electricity consumption of the kth user for:

[0030] If we represent the phase-by-phase electrical quantities of the transformer as a matrix W, then we have:

[0031] And assume that the line loss rate of the corresponding phase in each time period is Then the line loss matrix It can be represented as:

[0032] Based on the law of conservation of energy, we have:

[0033] When solving this problem using dynamic programming, we can treat the user's phase affiliation Y as the variable to be optimized. The algorithm searches for all possible combinations of user phases, with the goal of minimizing the following loss function:

[0034] Thus, having obtained the electricity consumption information of K users over L time periods and the phase-specific electricity consumption of each transformer, a dynamic programming algorithm can be used to solve for the phase sequence attribution relationship of the transformer area. However, within a fixed time period, users' electricity consumption behaviors are similar, making the models prone to confusion and limiting the recognition accuracy. Simply increasing the amount of data only increases the computational burden. Therefore, in order to improve data utilization and accuracy of results, this invention introduces a time consistency analysis strategy based on class ensemble learning to enhance the stability and reliability of identification across multiple time scales.

[0035] Specifically, let the length of the l-th time interval be T, and divide it into q time subsets. The topology assignment results are calculated separately. Ideally, each time subset should be consistent, but due to algorithm errors and data differences, the results may be inconsistent. Therefore, we take the intersection of the subset results as a high-confidence judgment; when q is sufficiently large, the intersection can be considered 100% accurate. Other structures not in the intersection are assigned probabilities based on their frequency of occurrence in each subset result, reflecting the likelihood that they are true structures.

[0036] To address the time-varying characteristics of user electricity consumption behavior, this invention employs an overlapping sliding window to pair sub-time series. Sampling improves data utilization efficiency. All sub-time series are set to a fixed length. If a user belongs to a certain phase in all sub-series (e.g., ... Figure 2 users in , If the frequency of a user exceeds a certain threshold, it can be determined that the user belongs to that phase 100%; other users are determined by the frequency of their subsequence. If the frequency exceeds the threshold, the user belongs to that phase; if the frequency is below the threshold, the user has low confidence and needs to be manually verified.

[0037] The method disclosed in this invention combines the principle of energy conservation with cross-time consistency analysis, requiring no additional hardware. Dynamic programming ensures the rationality of phase sequence assignment, and the introduction of time consistency and probabilistic judgment mechanisms significantly improves the stability and accuracy of the identification results. It can automatically identify results 100% without manual verification, providing efficient and accurate data support for intelligent operation and maintenance of distribution networks.

Claims

1. A multi-stage decision-making recursive optimization method for high-confidence radio station topology identification, used to solve the key topological unknown problem in power system state perception and analysis, characterized in that... The steps include: defining each phase of the three-phase power supply connected to the user as M different distribution zones. Each transformer substation contains K users. Let matrix X represent the electricity consumption of these K users over L time periods. Then: ,in: This represents the electricity consumption of K users within the current distribution area during the l-th time period. ; This describes the electricity consumption of the k-th user during the L-th time period. Let matrix Y represent the phase sequence attribution of K users, then we have: ,in, Let matrix W represent the phase-specific electrical quantities of the transformer, then we have: ,in, Let K be the electricity consumption of the k-th user in the m-th transformer area. Line loss matrix Represented as: ,in: Let m be the line loss rate of the m-th transformer station for the k-th user. Let be the total electricity consumption of the k-th user in the m-th transformer area; based on the law of conservation of energy, the following relationship is established: Treating user phase affiliation Y as the variable to be optimized, dynamic programming is used to solve the above equation. The algorithm searches for all possible combinations of user phases, with the goal of minimizing the following loss function: After obtaining the electricity consumption information of K users over L time periods and the phase-specific electricity on each transformer, a dynamic programming algorithm is used to solve the phase sequence attribution relationship of the transformer area.

2. The multi-stage decision-making recursive optimization method for high-confidence radio station topology identification as described in claim 1, characterized in that, The total electricity consumption of the kth user in the mth transformer area .

3. The multi-stage decision-making recursive optimization method for high-confidence radio station topology identification as described in claim 1, characterized in that, In the dynamic programming algorithm, the l-th time interval of length T is divided into q time subsets. The topological assignment results for each time subset are calculated separately. The intersection of the results of one subset is taken as the high confidence judgment, and the other structures not in the intersection are assigned probabilities based on their frequency of occurrence in each subset result.

4. The multi-stage decision-making recursive optimization method for high-confidence radio station topology identification as described in claim 3, characterized in that, In the dynamic programming algorithm: if the current user belongs to the same station area in all time subsets, then the current user is 100% considered to belong to the phase corresponding to that station area; otherwise, the current user is determined by the frequency of their belonging to each time subset: if the frequency of the current user's belonging to a certain station area in all time subsets exceeds the threshold, then the current user belongs to the phase corresponding to that station area; otherwise, the confidence level is low and manual verification is required.

5. The multi-stage decision-making recursive optimization method for high-confidence radio station topology identification as described in claim 3, characterized in that, An overlapping sliding window is used to sample the l-th time interval of length T, resulting in q time subsets. .

Images