A method and device for generating low voltage distribution network topology

Abstract

The present invention discloses a method and device for generating a low-voltage distribution network topology. The method includes: obtaining a time-series voltage data matrix according to electrical quantity data and re-assigning the abnormal acquisition data obtained by screening; running the t-SNE algorithm to obtain a low-dimensional voltage feature data set YT; running the DBSCAN algorithm to obtain all cluster sets C and a two-dimensional voltage feature clustering map; running the LLE algorithm to obtain a two-dimensional voltage feature map classified under the cluster label C; calculating and sorting the Euclidean distance relationship between the cluster centers and the master meter cluster in the feature map, and outputting the sorting result; generating a node adjacency matrix based on the topology recognition information obtained by the DBSCAN algorithm and the LLE algorithm; and visualizing the node adjacency matrix to generate a node connection topology map of the low-voltage distribution network. The present invention can visualize the topology information to generate a node topology map, providing information reference for various advanced applications such as power grid topology error correction and fault troubleshooting, and can be widely applied to the field of optimized operation and management of distribution networks.

Description

Technical Field

[0001] The present invention relates to the field of optimized operation and management of power distribution networks, and in particular to a method and device for generating a low-voltage power distribution network topology. Background Art

[0002] The low-voltage distribution network is directly connected to thousands of households, and the intelligent level of its operation, maintenance and management will directly affect the level of customer satisfaction. Correctly identifying the topological structure of the low-voltage area is of great significance for the power supply department to calculate the flow, update the change of switch status, analyze and judge the fault, conduct remote fee control, analyze the line loss and propose optimization strategies. However, with the development of the economy, the number of low-voltage power supply areas in various cities is increasing, the connection is becoming more and more chaotic, and there is even a phenomenon of not following the prescribed routing. With the increasing application of distributed power sources, controllable loads, electric vehicle access, and supply and demand response technology, the safe operation level and economy of the power grid have been improved. At the same time, the problem of frequent topological changes in urban low-voltage power supply areas has become increasingly prominent. Therefore, the topological structure recorded by the power supply department often has data loss, recording errors, etc., and the manual topological troubleshooting method is costly and inefficient. Exploring methods to efficiently, accurately and dynamically identify the power topological structure in the area and establish a reasonable, accurate and unified low-voltage topological model will help the power supply department to manage low-voltage area users, meet the power supply reliability requirements of the distribution network, and improve customer service quality.

[0003] At present, the mainstream topology structure identification methods can be divided into offline methods and online methods. The offline method requires manual hardware equipment to enter the site for testing and sorting, which consumes a lot of manpower and material resources, is costly, and has low efficiency and accuracy and cannot be automatically updated. The online method analyzes the information related to electricity consumption to identify the topology of the distribution network. It has the advantages of low cost and high real-time performance, such as the injection signal method, the data label method, and the data analysis method. But in general, most of the existing distribution network area topology identification methods cannot fully mine the electrical quantity data information collected by smart meters in actual projects. It is easy to identify simple structures and difficult to identify complex structures. When identifying the topology of a large-scale distribution network, there is still room for further improvement in accuracy, and the low-voltage distribution network topology identification technology that can be actually applied in engineering projects also needs to be further studied. Summary of the invention

[0004] In order to solve at least one of the technical problems existing in the prior art to a certain extent, an object of the present invention is to provide a method and device for generating a low-voltage power distribution network topology.

[0005] The technical solution adopted by the present invention is:

[0006] A method for generating a low voltage power distribution network topology comprises the following steps:

[0007] Obtain the electrical quantity data collected by the electric meters of each user in the low-voltage area, extract the collection information from the electrical quantity data, and generate the time-series voltage data matrix U according to the collection information; wherein the collection information includes the electric meter ID, voltage amplitude, and data collection time;

[0008] Screening abnormal collected data from the time series voltage data matrix, and reassigning the screened abnormal collected data;

[0009] Assign initial values ​​to the input parameters of the t-SNE algorithm, use the time series voltage data matrix U as the input data set of the t-SNE algorithm, and run the t-SNE algorithm to obtain the low-dimensional voltage feature data set Y T ;

[0010] Assign initial values ​​to the input parameters of the DBSCAN algorithm and convert the low-dimensional voltage feature data set Y T As the input data set of DBSCAN algorithm, run the DBSCAN algorithm to obtain all cluster sets C and two-dimensional voltage feature clustering diagram;

[0011] Assign initial values ​​to the input parameters of the LLE algorithm, use the time series voltage data matrix U as the input data set of the LLE algorithm, and run the LLE algorithm to obtain a two-dimensional voltage feature map under the cluster label C classification;

[0012] Calculate the Euclidean distance relationship between the cluster center and the total table cluster in the feature graph and sort them, output the sorting result, and represent the relative electrical distance relationship between different user branches and the total table;

[0013] Generate node adjacency matrix based on topological identification information obtained by DBSCAN algorithm and LLE algorithm;

[0014] The node adjacency matrix is ​​visualized to generate a node connection topology diagram of the low-voltage distribution network.

[0015] Furthermore, the expression of the timing voltage data matrix U is:

[0016]

[0017] In the formula, any voltage data U i,tj It is expressed as the meter i at t j The voltage amplitude measured at the time; m represents the number of users in all substations; n represents the number of voltage sampling points of the user collected by the meter in a certain period of time; any row vector U of the time series voltage data matrix U i Represents the time series voltage data of the same user meter at all times measured during the sampling period. Any column vector U tj It represents the voltage data of different users collected by each electricity meter at the same time.

[0018] Furthermore, the abnormal collected data obtained by screening is reassigned in the following way:

[0019]

[0020] In the formula, is a sampling time sequence t in the voltage matrix err Abnormal voltage data collected by the electric meter; and They represent the voltage sequence in the same time series that is closest to the abnormal data and whose sampling time is earlier than the time series t err and later than the timing t err Normal voltage data.

[0021] Furthermore, the input parameters of the t-SNE algorithm include n_components, perplexity, and learning_rate; wherein n_components represents the target dimension reduction, perplexity represents the perplexity, and learning_rate represents the learning rate;

[0022] The steps of the t-SNE algorithm are as follows:

[0023] A1. Convert the Euclidean distance of high-dimensional voltage data points into joint probability to express the correlation between each point. Use Gaussian distribution function to convert in high-dimensional space and calculate the conditional probability p j|i , p i|j And the joint probability distribution p ij ;

[0024] A2, using normal distribution N(0,10 -4 I) Randomly initialize the target low-dimensional dataset Y 0 ={y 1 ,y 2 ,...,y n};

[0025] A3. Use the t distribution function to transform in low-dimensional space and calculate the joint probability distribution q of data in low-dimensional space ij ;

[0026] A4. Make the probability distribution p ij =q ij , optimize the KL divergence between the two probability distributions and establish the objective function; iteratively calculate the gradient, and calculate the target low-dimensional data set Y according to the calculated gradient t Make updates;

[0027] A5. Determine whether the number of iterations reaches n_iter times; if so, execute step A6; if not, return to execute step A3;

[0028] A6. Get the low-dimensional feature dataset with the smallest KL divergence and use it as the t-SNE feature representation Y of the high-dimensional dataset T ={y 1 ,y 2 ,...,y n} and two-dimensional voltage characteristic diagram.

[0029] Furthermore, the conditional probability p j|i , p i|j And the joint probability distribution p ij The expression is as follows:

[0030]

[0031]

[0032]

[0033] In the formula, x i , x j , x k are any three high-dimensional vectors in the high-dimensional input data set; σ i and σ j x i and x j is the Gaussian function model variance of the Gaussian distribution center, which is determined by binary search through the input parameter perplexity; n is the number of high-dimensional vectors in the high-dimensional input data set;

[0034] The joint probability distribution of data in low-dimensional space q ij The expression is as follows:

[0035]

[0036] In the formula, y i ,y j ,y k ,y l Respectively represent the initialized or updated target low-dimensional dataset Y 0 or Y t Any 4 low-dimensional vectors in .

[0037] Furthermore, the iterative calculation of the gradient is based on the calculated gradient of the target low-dimensional data set Y t Updates include:

[0038] The gradient is iteratively calculated using the gradient descent method:

[0039]

[0040]

[0041] Where Y t-1 , Y t-2 Respectively represent Y t In the first two updated target low-dimensional datasets, η represents the learning rate, and α(t) represents the momentum threshold given by the algorithm.

[0042] Furthermore, the input parameters of the DBSCAN algorithm include ε and MinPts; wherein ε is the input neighborhood radius of each data sample object, and MinPts refers to the number of sample objects in the ε-neighborhood of the data sample;

[0043] The operation steps of the DBSCAN algorithm are as follows:

[0044] B1. From the low-dimensional voltage feature dataset Y T Select an unclassified core object as a seed, and obtain a set of data samples that have a density-reachable relationship with the core object as a cluster C. j ; Among them, the core object is defined as the object whose number of sample points in the ε-neighborhood is greater than or equal to MinPts; the density reachability relation is defined as for a certain sample set, given a string of sample points p 1 ,p 2 ,...,p n ,p=p 1 ,q=p n , the sample point p that meets the condition i In p i-1 ε-neighborhood, and p i-1 is a core object, then object q is density-reachable from object p;

[0045] B2, determine whether all core objects have categories, if yes, continue to execute step B3; otherwise, return to execute step B1;

[0046] B3. Mark a few abnormal sample points that are free outside the cluster as noise points. These noise points are not near any core object. The remaining normal sample points are divided into various clusters. Finally, all cluster sets C and two-dimensional voltage feature clustering diagrams are obtained.

[0047] Furthermore, the input parameters of the LLE algorithm include d, k, and C; where d represents the target dimension reduction, k represents the number of nearest neighbors, and C represents the cluster label;

[0048] The operation steps of the LLE algorithm are as follows:

[0049] C1. Find high-dimensional voltage data sample x based on Euclidean distance metric i The k nearest neighbors in the neighborhood Get high-dimensional voltage data sample xi The corresponding local covariance matrix;

[0050] C2. Obtain high-dimensional voltage data sample x i The corresponding weight coefficient vector;

[0051] C3, judging whether the condition is met: the local covariance matrix and weight coefficient vector corresponding to all high-dimensional voltage data samples are solved; if so, continue to step C4; otherwise, return to step C1;

[0052] C4, according to the weight coefficient vector W i A weight coefficient matrix W is formed, and a matrix M is calculated according to the weight coefficient matrix W;

[0053] C5. Calculate the first d+1 eigenvalues ​​of the matrix M and their corresponding eigenvectors, and then transform the matrix Y from the second eigenvector to the d+1th eigenvector into L ={y 2 ,y 3 ,...,y d+1} is used as the LLE feature representation of the high-dimensional data set, and the two-dimensional voltage feature map is output under the cluster label C classification.

[0054] Furthermore, the high-dimensional voltage data sample x is obtained by the following method: i The corresponding local covariance matrix:

[0055] Z i =(x i -x j )(x i -x j ) T

[0056] In the formula, x j Represents x i k nearest neighbors in the neighborhood Any one of the samples in

[0057] The high-dimensional voltage data sample x is obtained by the following method i The corresponding weight coefficient vector:

[0058]

[0059] In the formula, 1 k Represented as a k-dimensional all-1 vector;

[0060] The expression of matrix M is as follows:

[0061] M=(IW)(IW) T .

[0062] Another technical solution adopted by the present invention is:

[0063] A low voltage power distribution network topology generation device, comprising:

[0064] at least one processor;

[0065] at least one memory for storing at least one program;

[0066] When the at least one program is executed by the at least one processor, the at least one processor implements the above method.

[0067] The beneficial effect of the present invention is that the present invention uses an improved data dimension reduction clustering method to analyze the voltage spatiotemporal characteristics of each load node in the low-voltage network at a long time scale, obtains the possible connection relationship of each node, and then realizes the identification of node topology information in the low-voltage distribution network. Combined with graph theory knowledge, the topology information can be visualized to generate a node topology diagram, providing information reference for various advanced applications such as power grid topology error correction and troubleshooting. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the embodiments of the present invention or the drawings of related technical solutions in the prior art are introduced below. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0069] Figure 1 Schematic diagram of a typical topological structure of a low-voltage distribution station area in an embodiment of the present invention;

[0070] Figure 2 It is a low voltage distribution network simulation topology diagram in engineering embodiment 1 in an embodiment of the present invention;

[0071] Figure 3 is a low-voltage power distribution network topology prediction diagram generated by engineering embodiment 1 in an embodiment of the present invention;

[0072] Figure 4 It is a flow chart of a method for generating a low-voltage power distribution network topology in an embodiment of the present invention. DETAILED DESCRIPTION

[0073] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and are not to be construed as limitations of the present invention. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0074] In the description of the present invention, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., and orientations or positional relationships indicated are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0075] In the description of the present invention, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0076] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention based on the specific content of the technical solution.

[0077] In recent years, with the development and application of more and more intelligent terminal monitoring equipment, low-voltage distribution networks will have more visual monitoring interfaces, and large amounts of high-density user data information will be fed back to power grid companies through smart meters, providing a data source for the automatic identification of the topological structure of low-voltage distribution stations. The present invention uses user voltage data collected by the advanced measurement system (AMI) to propose a low-voltage distribution network topology generation method based on data dimensionality reduction clustering and graph theory knowledge. This method uses an improved data dimensionality reduction clustering method to analyze the voltage spatiotemporal characteristics of each load node in the low-voltage network over a long time scale, and derives the possible connection relationship between each node, thereby realizing the identification of node topology information in the low-voltage distribution network. Afterwards, combined with graph theory knowledge, the topological information can be visualized to generate a node topology diagram, providing information reference for various advanced applications such as power grid topology error correction and troubleshooting. At the same time, this method does not require additional manpower and material resources for equipment installation and maintenance, is more efficient and convenient, and has a high prospect for engineering application. Among them, Figure 1This is a schematic diagram of the typical topological structure of a low-voltage distribution station area.

[0078] like Figure 4 As shown, this embodiment provides a method for generating a low-voltage power distribution network topology based on data dimension reduction clustering and graph theory knowledge. The method is suitable for topology identification of a low-voltage power distribution network, and specifically includes the following steps:

[0079] S1. Obtain the complete electrical quantity data table collected by each user meter under the low-voltage area from the data acquisition center, extract the three key types of collection information (meter ID, voltage amplitude, data acquisition time), and generate the time series voltage data matrix U as shown in the following formula;

[0080]

[0081] Where: Any voltage data U i,tj It is expressed as the meter i at t j The voltage amplitude measured at the time; m represents the number of users in all substations; n represents the number of voltage sampling points of the user collected by the meter in a certain period of time. Any row vector U of the time series voltage data matrix U i Represents the time series voltage data of the same user meter at all times measured during the sampling period. Any column vector U tj It represents the voltage data of different users collected by each electricity meter at the same time.

[0082] S2. Filter abnormal collected data (zero voltage data and voltage data with voltage fluctuation greater than 10% of the normal value) from the voltage matrix U, and re-assign such data using the linear fitting method shown in the following formula to avoid affecting the recognition accuracy of the subsequent algorithm;

[0083]

[0084] Where: is a sampling time sequence t in the voltage matrix err Abnormal voltage data collected by the electric meter; and Respectively represent the distance abnormality collection data in the same time series voltage sequence The most recent and the sampling timing is earlier than t err and later than t err Normal voltage data.

[0085] S3. Assign initial values ​​to the t-SNE algorithm input parameters n_components, perplexity, learning_rate and n_iter, use the voltage matrix U as the input data set of the t-SNE algorithm, and execute the t-SNE algorithm. The specific execution steps include steps S4-S9.

[0086] Among them, n_components represents the target dimension of dimensionality reduction. For the visualization effect of the algorithm and the subsequent compatibility with the clustering algorithm, it is generally more appropriate to set it to 2; perplexity represents the perplexity. The size of this value reflects the different degrees of aggregation and dispersion of the dimensionality reduction data. Its value is generally set to the estimated cluster population size; learning_rate represents the learning rate, which affects the rate at which the cost function finds the optimal solution. It is generally set to 200; n_iter represents the number of iterations, which affects the convergence degree of the algorithm. It is generally set to 1000.

[0087] S4. First, the Euclidean distance of the high-dimensional voltage data points is converted into a joint probability to express the correlation between the points. The Gaussian distribution function is used for conversion in the high-dimensional space to calculate the conditional probability p j|i , p i|j And the joint probability distribution p ij , as shown below;

[0088]

[0089]

[0090]

[0091] Where: x i , x j , x k are any three high-dimensional vectors in the high-dimensional input data set; σ i and σ j x i and x j is the Gaussian function model variance of the Gaussian distribution center, which is determined by binary search through the input parameter perplexity; n is the number of high-dimensional vectors in the high-dimensional input data set.

[0092] S5. Using normal distribution N(0,10 -4 I) Randomly initialize the target low-dimensional dataset Y 0 ={y 1 ,y 2 ,...,y n};

[0093] S6. Use the t distribution function to transform in low-dimensional space and calculate the joint probability distribution q of data in low-dimensional space ij , as shown below;

[0094]

[0095] Where: y i ,y j ,yk ,y l Respectively represent the initialized or updated target low-dimensional dataset Y 0 or Y t Any 4 low-dimensional vectors in .

[0096] S7. Make p ij =q ij , optimize the KL divergence between the two probability distributions, establish the objective function, and iteratively calculate the gradient using the gradient descent method as shown in the following formula And for the target low-dimensional dataset Y t Make updates;

[0097]

[0098]

[0099] Where: Y t-1 , Y t-2 Respectively represent Y t In the first two updated target low-dimensional datasets, η represents the learning rate, and α(t) represents the momentum threshold given by the algorithm.

[0100] S8. Determine whether the condition is met: the number of iterations reaches n_iter times;

[0101] a. If satisfied, proceed to step S9;

[0102] b. If not satisfied, return to step S6;

[0103] S9. Get the low-dimensional feature dataset with the smallest KL divergence as the t-SNE feature representation Y of the high-dimensional dataset T ={y 1 ,y 2 ,...,y n} and two-dimensional voltage characteristic diagram.

[0104] S10. Assign initial values ​​to the DBSCAN algorithm input parameters ε and MinPts, and convert the low-dimensional voltage feature data set Y obtained in S7 into T The DBSCAN algorithm is executed as an input data set of the DBSCAN algorithm, and the specific execution steps include steps S11-S13.

[0105] Among them, ε is the input neighborhood radius of each data sample object, and MinPts refers to the number of sample objects in the ε-neighborhood of a data sample. The values ​​of the two parameters are generally determined by machine parameter adjustment.

[0106] S11. Start with Y TSelect an unclassified core object as a seed, and then find all the data sample sets that have a density-reachable relationship with the core object, which is a cluster C. j ;

[0107] Among them, the core object is defined as the object whose number of sample points in the ε-neighborhood is greater than or equal to MinPts; the density reachability relation is defined as for a certain sample set, given a string of sample points p 1 ,p 2 ,...,p n ,p=p 1 ,q=p n , the sample point p that meets the condition i In p i-1 ε-neighborhood, and p i-1 is a core object, then object q is density reachable from object p.

[0108] S12. Determine whether the condition is met: All core objects have categories

[0109] a. If satisfied, proceed to step S13;

[0110] b. If not satisfied, return to step S11;

[0111] S13. A few abnormal sample points that are free outside the cluster are marked as noise points. These points are not near any core object. The remaining normal sample points are divided into various clusters. Finally, all cluster sets C and two-dimensional voltage feature clustering diagrams are obtained, that is, the low-voltage area classification results and user phase classification results, which are output into a data table in the format of "meter ID + classification results".

[0112] S14. Assign initial values ​​to the LLE algorithm input parameters d, k and C, use the voltage matrix U obtained in S2 as the input data set of the LLE algorithm, and execute the LLE algorithm. The specific execution steps include steps S15-S19.

[0113] Among them, d represents the target dimension reduction. For the sake of algorithm visualization and subsequent compatibility with clustering algorithms, it is generally more appropriate to set it to 2; k represents the number of nearest neighbors, which reflects the number of linear representation samples of any data. The larger the k, the more accurate the representation. Its value is generally set to the total number of high-dimensional data samples - 1; C represents the clustering label obtained by S11.

[0114] S15. Find high-dimensional voltage data sample x based on Euclidean distance metric i The k nearest neighbors in the neighborhood Then use the following formula to find x i The corresponding local covariance matrix;

[0115] Zi =(x i -x j )(x i -x j ) T (9)

[0116] Where: x j Represents x i k nearest neighbors in the neighborhood Any one of the samples in .

[0117] S16. Use the following formula to find x i The corresponding weight coefficient vector;

[0118]

[0119] Where: 1 k Represented as a k-dimensional all-1 vector.

[0120] S17. Determine whether the condition is met: the local covariance matrix and weight coefficient vector corresponding to all high-dimensional voltage data samples are solved;

[0121] a. If satisfied, proceed to step S18;

[0122] b. If not satisfied, return to step S15;

[0123] S18. By weight coefficient vector W i The weight coefficient matrix W is formed, and the matrix M is calculated using the following formula;

[0124] M=(IW)(IW) T (11)

[0125] S19. Calculate the first d+1 eigenvalues ​​of the matrix M and their corresponding eigenvectors, and then transform the matrix Y spanned from the second eigenvector to the d+1th eigenvector L ={y 2 ,y 3 ,...,y d+1} as the LLE feature representation of the high-dimensional data set, and output the two-dimensional voltage feature map under the cluster label C classification;

[0126] S20. Calculate the Euclidean distance relationship between the cluster center and the total table cluster in the feature coordinate graph and sort them, output the sorting result, and indicate the relative electrical distance relationship between different user branches and the total table;

[0127] S21. Based on the topological identification information obtained in S13 and S19, a node adjacency matrix A is generated as shown below: ij ;

[0128]

[0129] S22. Visualize the node adjacency matrix to generate a low-voltage distribution network node connection topology diagram.

[0130] The above method is explained in detail below with reference to the accompanying drawings and specific embodiments.

[0131] Engineering Example 1:

[0132] Take two low-voltage distribution network simulation topology data samples as an example. Figure 2 As shown. A topology network represents a low-voltage area, and the node voltage represents the voltage measured by the user's smart meter. The specific topology parameters are: low-voltage distribution network simulation topology 1 contains 74 user nodes, topology 2 contains 82 user nodes, the voltage data collection length of each node is 30 days, the collection frequency is 15 minutes / time, and a total of 2880 voltage collection moments. Among them, Figure 2 (a) is a schematic diagram of simulation topology I. Figure 2 (b) Schematic diagram of simulation topology II.

[0133] Since the voltage data of some small branch nodes in the simulation topology are highly similar to those of the adjacent large branch nodes, it can be considered that merging a small branch into an adjacent large branch or classifying a node into an adjacent branch is also considered correct identification. The accuracy of the identification results is listed in Table 1. The low-voltage distribution network topology prediction diagram is shown in Figure 3 As shown. Among them, Figure 3 (a) is the topology prediction diagram of simulation topology I. Figure 3 (b) is the topology prediction diagram of simulation topology II.

[0134] Table 1 Analysis and recognition accuracy of simulation examples by tSNE-DBSCAN-LLE combined dimensionality reduction clustering method

[0135]

[0136] From the analysis of Table 1, it can be seen that for the low-voltage distribution network simulation topology data, that is, in the case of ideal data samples, the proposed recognition method can effectively identify the three types of topology information under the simulation topology, and generate a low-voltage distribution network prediction topology map based on the recognition information, thus realizing the preliminary prediction of the user topology of the low-voltage distribution area.

[0137] Engineering Example 2:

[0138] Taking the actual voltage data of users in three substations in Guangzhou, Guangdong, China as an example, the basic parameters of the three actual substations are shown in Table 2. The data collection length of each smart meter is 3 days, the data collection frequency is 1 minute / time, and the total number of moments is 4320.

[0139] Table 2 Basic parameters of actual substation area

[0140]

[0141] After on-site verification of the corresponding substations in this project embodiment, the accuracy of the recognition results is listed in Table 3.

[0142] Table 3 Comparison of the accuracy of area identification and phase identification results

[0143]

[0144] From the analysis of Table 3, it can be seen that for actual engineering data, the proposed identification method can effectively identify the low-voltage user-low-voltage area affiliation information and user phase information under the actual low-voltage area.

[0145] Since the data collection capabilities of the smart meters used in each region are different, in order to obtain the applicable scope of the proposed identification method, this embodiment also performs special processing on the input data set, and performs case tests on user voltage data with different sampling rates when the clock synchronization rate of the smart meter is 100% and the clock synchronization rate is 70%. The test results are listed in Tables 4 and 5, respectively.

[0146] Table 4 Test results of user voltage data with different sampling rates under 100% clock synchronization rate

[0147]

[0148] Table 5 Test results of user voltage data with different sampling rates at 70% clock synchronization rate

[0149]

[0150] The test results in Table 4 and Table 5 show that the proposed identification method can adapt to different data conditions, has certain effectiveness and advantages in solving the problem of low-voltage substation topology information identification, and can provide a reference for subsequent research in the field of low-voltage substation topology identification.

[0151] This embodiment also provides a low-voltage power distribution network topology generation device, including:

[0152] at least one processor;

[0153] at least one memory for storing at least one program;

[0154] When the at least one program is executed by the at least one processor, the at least one processor implements Figure 4 The method shown.

[0155] A low-voltage distribution network topology generation device of this embodiment can execute a low-voltage distribution network topology generation method provided by the method embodiment of the present invention, can execute any combination of implementation steps of the method embodiment, and has the corresponding functions and beneficial effects of the method.

[0156] The present application also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes Figure 4 The method shown.

[0157] This embodiment also provides a storage medium, which stores instructions or programs that can execute a low-voltage distribution network topology generation method provided by an embodiment of the method of the present invention. When the instructions or programs are run, any combination of implementation steps of the method embodiment can be executed, and the corresponding functions and beneficial effects of the method can be obtained.

[0158] In some selectable embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided by way of example, for the purpose of providing a more comprehensive understanding of technology. The disclosed method is not limited to the operation and logic flow presented herein. Selectable embodiments are expected, wherein the order of various operations is changed and the sub-operation of a part for which is described as a larger operation is performed independently.

[0159] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise specified, one or more of the functions and / or features described may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present invention. More specifically, in view of the properties, functions, and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional skills of the engineer. Therefore, those skilled in the art can implement the present invention set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0160] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0161] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0162] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0163] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0164] In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0165] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0166] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for generating low voltage distribution network topology, It is characterized in that The following steps are involved: Obtain the electrical quantity data collected by the electric meters of each user in the low-voltage area, extract the collection information from the electrical quantity data, and generate the time-series voltage data matrix U according to the collection information; wherein the collection information includes the electric meter ID, voltage amplitude, and data collection time; Filter abnormal data from the time series voltage data matrix, re-assign the abnormal data; assign initial values ​​to the input parameters of the t-SNE algorithm, use the time series voltage data matrix U as the input data set of the t-SNE algorithm, and run the t-SNE algorithm to obtain the low-dimensional voltage feature data set Y T ; Assign initial values ​​to the input parameters of the DBSCAN algorithm and convert the low-dimensional voltage feature data set Y T As the input data set of DBSCAN algorithm, run the DBSCAN algorithm to obtain all cluster sets C and two-dimensional voltage feature clustering diagram; Assign initial values ​​to the input parameters of the LLE algorithm, use the time series voltage data matrix U as the input data set of the LLE algorithm, and run the LLE algorithm to obtain a two-dimensional voltage feature map under the cluster label C classification; Calculate the Euclidean distance relationship between the cluster center and the total table cluster in the feature graph and sort them, output the sorting result, and represent the relative electrical distance relationship between different user branches and the total table; Generate node adjacency matrix based on topological identification information obtained by DBSCAN algorithm and LLE algorithm; Visualize the node adjacency matrix to generate a low-voltage distribution network node connection topology diagram; The input parameters of the LLE algorithm include d, k, and C; where d represents the target dimension reduction, k represents the number of nearest neighbors, and C represents the cluster label; The operation steps of the LLE algorithm are as follows: C1. Find high-dimensional voltage data sample x based on Euclidean distance metric i The k nearest neighbors in the neighborhood , get high-dimensional voltage data sample x i The corresponding local covariance matrix; C2. Obtain high-dimensional voltage data sample x i The corresponding weight coefficient vector; C3, judging whether the condition is met: the local covariance matrix and weight coefficient vector corresponding to all high-dimensional voltage data samples are solved; if so, continue to step C4; otherwise, return to step C1; C4, according to the weight coefficient vector W i A weight coefficient matrix W is formed, and a matrix M is calculated according to the weight coefficient matrix W; C5. Calculate the first d+1 eigenvalues ​​of the matrix M and their corresponding eigenvectors, and then transform the matrix Y from the second eigenvector to the d+1th eigenvector into L ={y 2 ,y 3 ,…,y d+1 } is used as the LLE feature representation of the high-dimensional data set, and the two-dimensional voltage feature map is output under the cluster label C classification.

2. A method for generating a low voltage distribution network topology according to claim 1, It is characterized in that The expression of the timing voltage data matrix U is: In the formula, any voltage data U i,tj It is expressed as the meter i at t j The voltage amplitude measured at the time; m represents the number of users in all substations; n represents the number of voltage sampling points of the user collected by the meter in a certain period of time; any row vector U of the time series voltage data matrix U i Represents the time series voltage data of the same user meter at all times measured during the sampling period. Any column vector U tj It represents the voltage data of different users collected by each electricity meter at the same time.

3. A method for generating a low voltage distribution network topology according to claim 1, It is characterized in that Reassign the abnormal collected data obtained through screening in the following ways: In the formula, is a sampling time sequence t in the voltage matrix err Abnormal voltage data collected by the electric meter; and They represent the voltage sequence in the same time series that is closest to the abnormal data and whose sampling time is earlier than the time series t err and later than the timing t err Normal voltage data.

4. A method for generating a low voltage distribution network topology according to claim 1, It is characterized in that The input parameters of the t-SNE algorithm include n_components, perplexity, and learning_rate; wherein n_components represents the target dimension reduction, perplexity represents the perplexity, and learning_rate represents the learning rate; The steps of the t-SNE algorithm are as follows: A1. Convert the Euclidean distance of high-dimensional voltage data points into joint probability to express the correlation between each point. Use Gaussian distribution function to convert in high-dimensional space and calculate the conditional probability p j|i , p i|j And the joint probability distribution p ij ; A2, using normal distribution N(0,10 -4 I) Randomly initialize the target low-dimensional dataset Y 0 ={y 1 ,y 2 ,...,y n }; A3. Use the t distribution function to transform in low-dimensional space and calculate the joint probability distribution q of data in low-dimensional space ij ; A4. Make the probability distribution p ij =q ij , optimize the KL divergence between the two probability distributions and establish the objective function; iteratively calculate the gradient, and calculate the target low-dimensional data set Y according to the calculated gradient t Make updates; A5. Determine whether the number of iterations reaches n_iter times; if so, execute step A6; if not, return to execute step A3; A6. Get the low-dimensional feature dataset with the smallest KL divergence and use it as the t-SNE feature representation Y of the high-dimensional dataset T ={y 1 ,y 2 ,...,y n } and two-dimensional voltage characteristic diagram.

5. A method for generating a low voltage power distribution network topology according to claim 4, It is characterized in that Conditional probability p j|i , p i|j And the joint probability distribution p ij The expression is as follows: In the formula, x i , x j , x k are any three high-dimensional vectors in the high-dimensional input data set; σ i and σ j x i and x j is the Gaussian function model variance of the Gaussian distribution center, which is determined by binary search through the input parameter perplexity; n is the number of high-dimensional vectors in the high-dimensional input data set; The joint probability distribution of data in low-dimensional space q ij The expression is as follows: In the formula, y i ,y j ,y k ,y l Respectively represent the initialized or updated target low-dimensional dataset Y 0 or Y t Any 4 low-dimensional vectors in .

6. A method for generating a low voltage distribution network topology according to claim 4, It is characterized in that The iterative calculation of the gradient is based on the calculated gradient of the target low-dimensional data set Y t Updates include: The gradient is iteratively calculated using the gradient descent method: Where Y t-1 , Y t-2 Respectively represent Y t In the first two updated target low-dimensional datasets, η represents the learning rate, and α(t) represents the momentum threshold given by the algorithm.

7. A method for generating a low voltage power distribution network topology according to claim 1, It is characterized in that The input parameters of the DBSCAN algorithm include ε and MinPts; where ε is the input neighborhood radius of each data sample object. MinPts refers to the number of sample objects in the ε-neighborhood of the data sample; The operation steps of the DBSCAN algorithm are as follows: B1. From the low-dimensional voltage feature dataset Y T Select an unclassified core object as a seed, and obtain a set of data samples that have a density-reachable relationship with the core object as a cluster C. j ; Among them, the core object is defined as the object whose number of sample points in the ε-neighborhood is greater than or equal to MinPts; the density reachability relation is defined as for a certain sample set, given a string of sample points p 1 ,p 2 ,...,p n ,p=p 1 ,q=p n , satisfying the condition that the sample point pi is in pi- 1 ε-neighborhood, and pi- 1 is a core object, then object q is density-reachable from object p; B2. Determine whether all core objects have categories. If so, continue to step B3; otherwise, return to step B1； B3. Mark a few abnormal sample points that are free outside the cluster as noise points. These noise points are not near any core object. The remaining normal sample points are divided into various clusters. Finally, all cluster sets C and two-dimensional voltage feature clustering diagrams are obtained.

8. A method for generating a low voltage distribution network topology according to claim 1, It is characterized in that The high-dimensional voltage data sample x is obtained by the following method i The corresponding local covariance matrix: Z i =(x i -x j )(x i -x j ) T In the formula, x j Represents x i k nearest neighbors in the neighborhood Any one of the samples in The weight coefficient vector corresponding to the high-dimensional voltage data sample xi is obtained by the following method: In the formula, 1 k Represented as a k-dimensional all-1 vector; The expression of matrix M is as follows: M=(I-W)(I-W) T 。 9. A low voltage distribution network topology generation device, It is characterized in that include: at least one processor; at least one memory for storing at least one program; When the at least one program is executed by the at least one processor, the at least one processor implements the method according to any one of claims 1 to 8.

Images