Topology identification method based on intelligent measurement data of low-voltage transformer area

Abstract

The invention discloses a topology identification method based on intelligent measurement data of a low-voltage transformer area. According to the technical scheme, the topology identification method comprises the steps that 1, a master station issues a timing command to an intelligent measurement terminal and an electric meter; step 2, the master station issues a transformer area electric meter file to the intelligent measurement terminal, configures an acquisition task and issues the acquisition task to the intelligent measurement terminal; step 3, establishing an electric quantity sum linear regression model based on a Lasso algorithm by using daily frozen electric quantity data, and converting the identification problem of the transformer area user-transformer relation into regression coefficient solving; step 4, on the basis of user change relation identification, performing dimension reduction on historical voltage data by using a t-SNE algorithm to improve the identification efficiency of the model; and step 5, clustering the voltage data after dimension reduction by using agglomerated hierarchical clustering, identifying a user phase and a meter box to which the user belongs, and completing identification of a topological structure. According to the method, the power grid topology information of the transformer area can be automatically identified, the efficiency and the accuracy of topology identification are improved, and the method has far-reaching significance for construction of an intelligent power grid.

Description

Technical Field

[0001] This invention relates to the field of low-voltage smart distribution networks, and in particular to a method for identifying the topology of low-voltage distribution networks based on big data analysis. Background Technology

[0002] Currently, the main methods for identifying transformer substation topology relationships are the characteristic signal method and the power frequency zero-crossing method. The characteristic signal method requires additional equipment and is affected by the communication method, has a long identification cycle, and is easily affected by the load of the substation. The power frequency zero-crossing method has high requirements for clock synchronization and detection accuracy, and requires the establishment of a carrier communication network in the substation area, which may cause noise to the power grid and affect stability. Therefore, it is necessary to find a reliable, economical method with minimal impact on the power grid to identify the relationship between households and transformers in a transformer substation area.

[0003] With the widespread application of smart metering terminals and centralized meter reading technology, massive amounts of electricity consumption data from transformer and user sides have been provided. Traditional methods for identifying transformers based on voltage sequence similarity are prone to misjudgment due to similar voltage curves in adjacent transformer areas. Methods using symmetric elimination to reduce the order of the adjacency matrix and applying symmetry to the connection matrix calculation reduces the computational load for topology identification, but require high-quality measurement data. A physical model based on power equations and power flow calculation methods, incorporating linear regression and Jacobian matrices for iterative solutions, achieves topology identification based on measurement data. However, this requires measurement devices at each node and demands high-quality measurement data, thus limiting its engineering applicability to some extent. Summary of the Invention

[0004] This invention addresses the shortcomings and defects of existing technologies by providing a topology identification method based on intelligent measurement data of low-voltage distribution areas. This method can automatically identify and correct the power grid topology information of distribution areas, realize refined management of the distribution network, improve the efficiency and accuracy of topology identification, and has profound significance for the construction of smart grids.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A topology identification method based on intelligent measurement data of low-voltage distribution areas includes the following steps:

[0007] Step 1: First, the metering automation master station sends instructions to the intelligent metering terminal to calibrate the time of all meters in the distribution area to ensure the synchronization of the collected data.

[0008] Step 2: Then, the metering automation master station sends the meter files of the distribution area to the intelligent measurement terminal, configures the data collection task, and sends it to the intelligent measurement terminal.

[0009] Step 3: Using the daily frozen electricity data from the smart metering terminal in the transformer area and the user's smart meter, establish an electricity summation linear regression model based on the Lasso algorithm, transforming the problem of identifying the relationship between the transformer area and the user into solving the regression coefficients.

[0010] Step 4: Based on the identification of household-transformer relationships, the t-SNE algorithm is used to reduce the dimensionality of historical voltage data, thereby reducing the identification error caused by noise and redundant information and improving the model's identification efficiency.

[0011] Step 5: Cluster the dimensionality-reduced voltage data using agglomerative hierarchical clustering to identify user phases. Based on the identified phases, determine the user's meter box according to voltage data characteristics to complete the topology identification.

[0012] Further, in step 3, based on the daily frozen electricity data from the smart metering terminal (regional master meter) and smart meters, a Lasso regression model is used to analyze the relationship between the regional transformer area and the household transformer, and then the ownership of the household transformer is determined according to the magnitude of the regression coefficient. Considering line losses, a regional transformer-household transformer electricity relationship model is established:

[0013] Y = Xβ + ε

[0014] In the formula, Y is the column vector of total electricity consumption of the transformer area, X is the user electricity consumption matrix, β is the column vector of regression coefficients, ε is the column vector of errors, and each error variable follows a Gaussian distribution.

[0015] Constructing the objective function for Lasso regression, an algorithm for feature selection and sparse modeling, involves using an L1 regularization term to optimize the loss function, thereby reducing the coefficients of some features to zero and achieving feature selection. Lasso regression effectively handles high-dimensional data and multicollinearity. Its goal is to minimize the sum of the loss function and the regularization term, resulting in a sparse model. The specific optimization objective is as follows:

[0016]

[0017] In the formula, λ≥0 is the penalty coefficient, and its value affects the number of non-zero coefficients in the regression coefficient vector.

[0018] Furthermore, in step 4, to improve recognition accuracy, a large amount of data is required. However, excessive data volume can increase running time, thus affecting recognition efficiency. Therefore, this invention employs the t-SNE dimensionality reduction algorithm to preprocess the data, mapping high-dimensional data to a low-dimensional space. The t-SNE algorithm represents the similarity between voltage samples by converting Euclidean distance into conditional probabilities. This method can effectively reduce data dimensionality and improve recognition efficiency. Let the initial voltage dataset be U, for any two voltage samples U... i and U j , using p j|i The one-sided probability between sample points is defined as:

[0019]

[0020] In the formula, p j|i Represents voltage data U j In voltage data U i The probability of occurrence, σ i Based on voltage sample U i The standard deviation of a Gaussian distribution centered at .

[0021] Let the voltage dataset after dimensionality reduction be V. Calculate V using a t-distribution in the low-dimensional space. i and V j The probability distribution q between ij The formula is expressed as:

[0022]

[0023] To analyze the information loss between two probability distributions of high-dimensional and low-dimensional data by calculating KL divergence, the ultimate optimization objective is to minimize the KL divergence, i.e.:

[0024]

[0025] In the formula, P and Q are both M*N matrices, and their elements are p, p, and Q respectively. ij and q ij The smaller C is, the more consistent the probability distribution of the user voltage dataset is before and after dimensionality reduction, and the less information loss there is.

[0026] Further, in step 5, a clustering hierarchical clustering method is adopted, setting the number of clusters to 3, and performing hierarchical clustering on the low-dimensional voltage dataset. This allows single-phase users with similar phase voltages to the corresponding transformer area to be clustered into the same phase. Given the user's meter connection phase information and the number of meter boxes k, the user voltage dataset is subjected to secondary hierarchical clustering according to each phase. Setting the number of clusters to k, users belonging to the same category are grouped into the same meter box. Simultaneously, users with higher average voltages in different clusters are considered to be located upstream of the line. Before clustering, each object is treated as an independent cluster, and the similarity between objects is calculated according to a similarity metric. Subsequently, objects are gradually merged to form new clusters in descending order of similarity, repeating this merging process until the required number of clusters is reached. Euclidean distance is chosen as the similarity metric to reflect the similarity between voltage samples. The Euclidean distance between the feature vectors of the voltage data of two users after dimensionality reduction is:

[0027]

[0028] Given the varying definitions of similarity between different clusters, this invention chooses the average chain method as the basis for user phase characteristic clustering to ensure the flexibility and accuracy of the clustering model. The formula for the inter-cluster distance between two different clusters using the average chain method is as follows:

[0029]

[0030] In the formula, ||xz|| is the distance between two voltage data points x and z, and |C i |and|C j |These are C-classes i and C j The number of voltage data points.

[0031] Compared with existing technologies, the beneficial technical effects of this invention are as follows: This invention utilizes intelligent measurement data information and proposes an automatic identification method for distribution network topology based on the Lasso regression model, t-SNE algorithm, and agglomerative hierarchical clustering algorithm. This method can automatically identify and correct the distribution network topology information, realize refined management of the distribution network, and improve the efficiency and accuracy of topology identification. Attached Figure Description

[0032] Figure 1 This is a flowchart of a topology identification method based on intelligent measurement data of low-voltage distribution areas according to the present invention.

[0033] Figure 2 This is a flowchart of the household change relationship identification based on the Lasso algorithm of this invention.

[0034] Figure 3This is a flowchart of the dimensionality reduction process for historical voltage data based on the t-SNE algorithm of this invention.

[0035] Figure 4 This is a flowchart of the phase and meter box identification based on agglomerative hierarchical clustering of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and do not limit the scope of the invention.

[0037] A topology identification method based on intelligent measurement data of low-voltage distribution areas includes the following steps:

[0038] Step 1: First, the metering automation master station sends instructions to the intelligent metering terminal to calibrate the time of all meters in the distribution area to ensure the synchronization of the collected data.

[0039] Step 2: Then, the metering automation master station sends the meter files of the distribution area to the intelligent measurement terminal, configures the data collection task, and sends it to the intelligent measurement terminal.

[0040] Step 3: Using the daily frozen electricity data from the smart metering terminals in the transformer substations and user smart meters, establish a linear regression model based on the Lasso algorithm to transform the problem of identifying the relationship between transformer substations and users into solving for the regression coefficients. The specific process is as follows: Figure 1 As shown, firstly, the daily frozen electricity consumption data of the smart metering terminal (main meter of the distribution area) and user smart meters are obtained. The daily electricity consumption data of the smart metering terminal to be identified and n low-voltage user meters over m days are then converted into matrix form. The electricity consumption data of the smart metering terminal constitutes an m-dimensional column vector Y, while the electricity consumption data of the low-voltage user meters constitutes an m×n matrix X. Next, an identification model is established, the data is imported, and 10-fold cross-validation is used to solve for the optimal penalty coefficient. The coordinate descent method is then used to solve for the regression coefficient. If the regression coefficient of a user's electricity consumption is close to 1, it indicates that the user belongs to that distribution area; if the regression coefficient is close to 0, it indicates that the user does not belong to that distribution area.

[0041] Step 4: Based on the identification of household-transformer relationships, the t-SNE algorithm is used to reduce the dimensionality of historical voltage data, thereby reducing identification errors caused by noise and redundant information and improving the model's identification efficiency. The specific process is as follows: Figure 2 As shown, the voltage datasets of the branch line side and single-phase users of the station are obtained. The perplexity, number of iterations and the dimension to be reduced are set. The probability distribution is calculated to generate the dimension-reduced voltage dataset. The probability density of the low-dimensional dataset is calculated. The process is repeated S times to obtain the final dimension-reduced voltage dataset.

[0042] Step 5: Cluster the dimensionality-reduced voltage data using agglomerative hierarchical clustering to identify user phases. Based on the phase identification, determine the user's meter box according to voltage data characteristics, completing the topology identification; the specific process is as follows... Figure 3 As shown, firstly, a reduced-dimensional voltage dataset is input, and the number of phase clusters is set to 3. Hierarchical clustering is performed on the low-dimensional data to obtain phase identification results. Then, depending on the specific situation, the number of meter boxes is set to k, and secondary hierarchical clustering is performed separately for each phase. Users of the same category are grouped into the same meter box, and in different clusters, users with higher average voltage are considered to be located upstream of the power line. Finally, the transformer substation topology is obtained.

[0043] The above embodiments are descriptions of specific implementations of the present invention, and not limitations thereof. Those skilled in the art can make various modifications and changes without departing from the spirit and scope of the present invention to obtain corresponding equivalent technical solutions. Therefore, all equivalent technical solutions should be included in the patent protection scope of the present invention.

Claims

1. A topology identification method based on intelligent measurement data of low-voltage distribution areas, characterized in that, This includes applications involving metering automation master stations, smart measurement terminals, smart energy meters, and communication channels, and includes the following steps: Step 1: First, the metering automation master station sends instructions to the intelligent metering terminal to calibrate the time of all meters in the distribution area to ensure the synchronization of the collected data; Step 2: Then, the metering automation master station sends the electricity meter files of the distribution area to the intelligent metering terminal, configures the data collection task, and sends it to the intelligent metering terminal. Step 3: Using the daily frozen electricity data from the smart metering terminal in the transformer area and the user's smart meter, establish an electricity summation linear regression model based on the Lasso algorithm, transforming the problem of identifying the relationship between the transformer area and the user into solving the regression coefficients. Step 4: Based on the identification of household-transformer relationships, the t-SNE algorithm is used to reduce the dimensionality of historical voltage data, thereby reducing the identification error caused by noise and redundant information and improving the model's identification efficiency. Step 5: Cluster the dimensionality-reduced voltage data using agglomerative hierarchical clustering to identify user phases; based on the identified phases, identify the meter box to which the user belongs according to the voltage data characteristics, thus completing the identification of the topology.

2. The topology identification method based on intelligent measurement data of low-voltage distribution areas according to claim 1, characterized in that, The metering automation master station is an integrated application platform capable of uniformly collecting, monitoring, statistically analyzing, and managing electrical energy data from substations, dedicated transformers, public transformers, and low-voltage users. The intelligent measurement terminal simultaneously supports five functional modules. By changing modules and downloading an application APP, it can realize terminal functions such as load control, distribution transformer, concentrator, and TTU. It can collect various data, including electricity, current, voltage, and power, on a timed and random basis, and collect real-time data, curve data, and daily data. The intelligent measurement terminal connects to the master station via 4G to exchange data. The communication channel enables bidirectional communication between the energy meter and the intelligent measurement terminal, and can be divided into RS485 bus communication, dual-mode communication, and power line carrier communication.

3. The topology identification method based on intelligent measurement data of low-voltage distribution areas according to claim 1, characterized in that, In step 1, the master station issues a time synchronization command, and the terminal synchronizes the time of all electricity meters in the distribution area by broadcasting the time synchronization. After the time synchronization, the master station calls the real-time time of the terminal and the electricity meters to ensure clock synchronization.

4. The topology identification method based on intelligent measurement data of low-voltage distribution areas according to claim 1, characterized in that, The data collection tasks issued by the main station in step 2 include the hourly voltage curve of the electricity meter and the daily frozen electricity data.

5. The topology identification method based on intelligent measurement data of low-voltage distribution areas according to claim 1, characterized in that, In step 3, based on the daily frozen electricity data from the smart metering terminal (regional master meter) and smart meters, a Lasso regression model is used to analyze the relationship between the regional transformer area and the household transformer. Then, the ownership of the household transformer is determined based on the magnitude of the regression coefficient. Considering line loss, a regional transformer-household transformer electricity relationship model is established: Y = Xβ + ε In the formula, Y is the column vector of total electricity consumption of the transformer area, X is the user electricity consumption matrix, β is the column vector of regression coefficients, ε is the column vector of error, and each error variable follows a Gaussian distribution; Construct the objective function for Lasso regression, with the following optimization objective: In the formula, λ≥0 is the penalty coefficient, and its value affects the number of non-zero coefficients in the regression coefficient vector.

6. The topology identification method based on intelligent measurement data of low-voltage distribution areas according to claim 1, characterized in that, In step 4, to address the increased computation time of the algorithm due to large-scale data, the voltage sequence features are reduced in dimensionality using the t-SNE algorithm before user phase identification.

7. The topology identification method based on intelligent measurement data of low-voltage distribution areas according to claim 1, characterized in that, Step 5 employs agglomerative hierarchical clustering algorithm to automatically identify user phases and meter boxes; Euclidean distance is selected as the similarity metric to reflect the similarity between voltage samples; the formula for the Euclidean distance between N-dimensional spatial points is: The similarity of clusters is determined by calculating the distance between different clusters; the formula for the distance between clusters is: In the formula, |xz| represents the distance between two voltage data points x and z, and |C| represents the distance between them. i |and|C j |These are C-classes i and C j The number of voltage data points.

Images