Low-voltage distribution network user-transformer relation identification method based on power balance and knowledge graph and computer medium
By combining power balance and knowledge graph-based methods with Gaussian mixture models and correlation convolution optimization, the accuracy and efficiency issues of household transformer relationship identification in low-voltage distribution networks were solved, achieving efficient and accurate household transformer relationship identification.
Patent Information
- Application Number
- CN202411465405.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-21
- Publication Date
- 2025-02-25
AI Technical Summary
Existing technologies are insufficient for efficiently and accurately identifying the relationships between customers and transformers in low-voltage distribution networks, resulting in difficulties in identification, poor accuracy, waste of resources, and significant investment.
A power balance and knowledge graph-based approach is adopted. By collecting the location and voltage information of end users, cluster analysis is performed by combining Gaussian mixture model and correlation convolution optimization model, and topology correction is performed by combining knowledge graph, so as to achieve accurate identification of the relationship between households and transformers.
It improves the accuracy and efficiency of identifying the relationship between households and transformers, reduces resource waste and time costs, and accurately reflects the actual topology of the low-voltage distribution network.
Smart Images
Figure CN119513633A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of household-transformer relationship identification in a low-voltage distribution network, and in particular to a household-transformer relationship identification method and a computer medium for a low-voltage distribution network based on power balance and knowledge graph. Background Art
[0002] The topological structure of the distribution network is the basis for power flow calculation, state perception and safety control of the distribution network. With the large-scale grid connection of renewable distributed power sources, the topological structure and operating characteristics of the distribution network are becoming increasingly complex and diverse. The occurrence of uncertain events such as system failures and equipment maintenance causes the intermittent opening and closing of the section switches on the distribution network tie lines. Due to the limited number of sensors and the inability to meet the real-time requirements of signal transmission, it is impossible to obtain an accurate distribution network topological structure. Therefore, accurate identification of the distribution network topological structure has become an important guarantee for the safe and stable operation of the distribution network.
[0003] The relationship between household transformers in low-voltage distribution networks is an important part of the topological structure of distribution networks. Correctly identifying the relationship between household transformers can improve the intelligent management level of distribution networks. In the actual operation of distribution networks, there are many problems such as the large number of low-voltage distribution stations and low-voltage users, and the lack of measuring devices in the distribution system, which leads to unclear household transformer relationships and difficult identification. At present, the identification of household transformer relationships mainly adopts manual on-site inspections, additional measuring equipment, and data-driven methods. However, manual inspections waste a lot of manpower and material resources, increase the investment in measuring equipment, and the data-driven method requires a large amount of complete measurement data. These problems lead to poor results and poor accuracy of existing household transformer relationship identification methods.
[0004] Therefore, a new technical solution is urgently needed to solve the technical problem of how to efficiently and accurately identify the relationship between household transformers in the distribution network. Summary of the invention
[0005] The present invention provides a method and computer medium for identifying the relationship between household transformers in a low-voltage distribution network based on power balance and knowledge graph, so as to solve the technical problem of how to efficiently and accurately identify the relationship between household transformers in a distribution network.
[0006] To achieve the above object, the present invention provides a method for identifying the relationship between household transformers in a low-voltage distribution network based on power balance and knowledge graph, comprising the following steps:
[0007] S1. Collect first information and second information; the first information includes the location information and voltage information of the terminal user; the second information includes the distribution transformer data; and cluster the initial addresses of neighboring terminal users according to the first information.
[0008] S2. Perform cluster analysis on the terminal user voltage time series data based on the initial address clustering and the second information based on a Gaussian mixture model to obtain a preliminary classification result of the terminal users.
[0009] S3. Based on the preliminary division results and the power balance principle, an associated convolution optimization model is constructed and iteratively solved to obtain the preliminary household-to-substation relationship in the low-voltage distribution system.
[0010] S4. Based on the knowledge graph, the preliminary household-to-household relationship in the low-voltage distribution system is modified to obtain the household-to-household relationship in the low-voltage distribution network and complete the household-to-household relationship identification.
[0011] Preferably, collecting the first information and the second information in S1 includes:
[0012] Assume that there are m distribution transformers and n end users in the area to be identified. The set M of distribution transformers and the set N of end users can be expressed by the following relationship:
[0013] M={1,2,...,m}
[0014] N = {1, 2, ..., n}
[0015] The location information and voltage information of the terminal users in the area to be identified are collected to obtain the first information; and the distribution transformer data are collected to obtain the second information.
[0016] The location information includes the user's address information in a preset format, and the preset format includes four gradually refined parts: cell information, building information, unit information, and room information; the voltage information includes the user's power information, active power information, and voltage information.
[0017] Preferably, clustering the initial addresses of neighboring terminal users according to the first information in S1 includes:
[0018] The terminal users whose location information, namely, the cell information, the building information, and the unit information, are the same in the first information are screened and clustered to obtain a user set U.
[0019] Calculate the voltage information of the end users in the user set U:
[0020] According to the room information in the user set U, the terminal users on the floor s are filtered and clustered to obtain the neighbor user set Us.
[0021] Use a central user node to replace the neighbor user set Us and get the power of the central user node Active Power And voltage data sequence Defining power and active power They are the total electricity and active power of all end users in the set Us:
[0022]
[0023] in, Indicates the power of the user in Us; Indicates the active power of the user in Us; and They respectively represent the electricity data and active power data of the central user node Us at the tth moment, t=1, 2, ..., W; T represents transposition.
[0024] The original voltage data of the terminal user in the first information is analyzed by the Z-Score method. To standardize:
[0025]
[0026] in, represents the standardized voltage value of the end user at time t; represents the mean voltage of the terminal users at time t; Represents the voltage standard deviation of the terminal user at time i.
[0027] Define the voltage of the central user node of Us is the mean voltage of all users in Us:
[0028]
[0029] Among them, k is the number of terminal users in Us; l∈US is the lth terminal user in Us.
[0030] Based on this, the voltage data sequence is obtained
[0031]
[0032] in, is the voltage data of the central user at the tth moment.
[0033] Preferably, S2 comprises:
[0034] S21. Initialize model parameters:
[0035] Define the number of clusters k cluster , k cluster The value is equal to the number of distribution transformers m; define the voltage data of the central user node of Us for The feature vector in W-dimensional sample space Gaussian mixture distribution Defined as:
[0036]
[0037] Among them, μ is the mean of the W-dimensional vector; θ is the W×W-dimensional covariance matrix; η j is the Gaussian mixing coefficient, which indicates the probability of selecting the jth mixing component, and its value is greater than 0; exp(·) is an exponential function.
[0038] By k cluster Gaussian mixture components, μ j and θ j is the mean vector and covariance matrix of the jth Gaussian component; at the beginning of clustering, the model parameters {(η j , μ j ,θ j )|1≤j≤k cluster}.
[0039] S22. Update model parameters:
[0040] In each iteration, the posterior probability Ψ of the sample belonging to each Gaussian component is calculated based on the current parameters. t，j , and then the model parameter η j , μ j ,θ j Perform iterative updates.
[0041] S23, according to Gaussian mixture distribution, Divide into k cluster Cluster sample The cluster The judgment formula can be expressed as:
[0042] γ i = argmaxΨ t，j , j∈{1,…,k cluster}
[0043]
[0044] That is, the preliminary division result of the user is obtained.
[0045] Preferably, S22 includes:
[0046] Set the random variable g j ∈{1, 2, ..., k cluster} indicates generating samples Gaussian mixture components.
[0047] According to Bayes' theorem, we can get:
[0048]
[0049] Among them, t，j For sample The posterior probability generated by the jth Gaussian mixture component; μ r and θ r is the mean vector and covariance matrix of the rth Gaussian component; r is the random variable g j The rth element in .
[0050] To solve the model parameters {(η j , μ j ,θ j )|1≤j≤k cluster}, It can be further expressed as:
[0051]
[0052] The model parameters are iteratively optimized using the maximum expectation algorithm to achieve continuous updating. j , μ j ,θ j )|1≤j≤k cluster}Take the maximum value, From this we can get:
[0053]
[0054] Then we can get μ j :
[0055]
[0056] make Then we can find θ j :
[0057]
[0058] For the mixing coefficient η j , must satisfy η j ≥0, Its Lagrangian form is:
[0059]
[0060] By taking the derivative of the mixing coefficient and setting the inverse to 0, we can obtain η j :
[0061]
[0062] Preferably, S3 includes:
[0063] Define the binary association variable x, then the connection relationship between the distribution transformer node j and the terminal user node i in the preliminary division result can be expressed as:
[0064]
[0065] Among them, Ω j is the set of users within the power supply range of low-voltage distribution transformer j and having electrical connectivity with distribution transformer j; i，j is the connection relationship between i and j. When end user i is within the power supply range of distribution transformer j and has an electrical connection with the distribution transformer, x j，i =1; On the contrary, when there is no electrical connection between the end user i and the distribution transformer j, x j , i =0.
[0066] In a low-voltage distribution transformer system including m distribution transformers and n end users, the sets of distribution transformers and end users are M = [1, 2, ..., j, ..., m] and N = [1, 2, ..., i, ..., n] respectively; for any low-voltage distribution transformer j∈M, define the association vector X j Describe the connectivity relationship between distribution transformer j and all users:
[0067] X j =[x j，1 , x j，2 , ..., x j，n ]
[0068] The association vector combination of m distribution transformers together constitutes a 0-1 adjacency matrix X that represents the connectivity relationship between all distribution transformers and end users in the low-voltage distribution system, that is, the connection relationship matrix:
[0069] X=[X 1 , X 2 , ..., X m ] T
[0070] There is a unique mapping relationship between distribution transformers and end users, then:
[0071]
[0072] Among them, x j，i Represents the element at row j and column i in X.
[0073] The associative convolution model describing the connection relationship between distribution transformer j and end users can be expressed as:
[0074]
[0075] Among them, * represents the associated convolution operation. The convolution kernel slides on the input data matrix along the time axis with a sliding step of 1. The weighted sum of the coverage area is calculated each time, and the output data matrix is finally obtained; z N and p N Input matrix for electricity and power data; E M Represents the power output matrix; P M represents the power output matrix; e N represents the power input matrix; p N represents the power input matrix; Indicates E M The element in row t of ; Indicates P M The element in row t of ; Indicates E M The jth element in the tth row; Indicates P M The jth element in the tth row; Indicates e N The i-th element in the t-th row; Indicates p N The ith element in the tth row.
[0076] The household-to-household relationship based on energy supply and demand balance, i.e. power balance, can be described as:
[0077] E M =e N *X+ξ
[0078] X T H=1 n
[0079] Where ξ is the error matrix, which is generated by the measurement, noise, and line loss during the energy transmission process of distribution transformer j; H is a matrix composed of m matrices D, H = [D, D, ..., D] T , D is the n×n dimensional identity matrix; 1 n is a 1×n dimensional matrix whose elements are all 1.
[0080] The household change relationship problem is transformed into an optimization problem, by solving the unknown binary variables x in the connection relationship matrix j，i Finally, the physical connection relationship between the distribution transformer and the end user is determined. In order to increase the solvability of the problem, the Softmax function is introduced to the binary variable x j，i Relax to get x′ j，i :
[0081]
[0082] X = [x′ j，i ]m×n
[0083]
[0084] Among them, e () It means to find the exponentiation with e as the base; h=1 is the starting number for the sum.
[0085] Considering the minimum sum of squares of the error between the energy supplied by the distribution transformer and the total energy consumption of the end user, the objective function is defined as:
[0086]
[0087] in, Represents the tth element of the power input matrix.
[0088] Solving minJ 1 , we get the matrix X, that is, we get the preliminary household-to-household relationship in the low-voltage distribution system.
[0089] Preferably, S4 includes:
[0090] S41. Obtain the location information of the terminal users in the area to be identified based on the knowledge graph, cluster the terminal users with the same cell information and building information, and obtain the terminal user set Ux of the Bth building.
[0091] S42. Count the number of distribution transformers N in the Bth building based on the knowledge graph B :
[0092]
[0093] S43, according to the capacity SN of the distribution transformer and the rated power factor of the distribution transformer The number of terminal users in building B is N u and the typical load PN of low-voltage users, and obtain the number of transformers N in the building that meets the relevant national standards. D :
[0094]
[0095] S44. Solve for N B With N D The difference between B -N D ≥1, indicating that there is an error in the household change relationship of Building B, and preliminary confirmation requires on-site verification.
[0096] S45. After preliminary confirmation that field verification is required, use the elements in the knowledge graph to compare with the preliminary household-to-household relationship; if the topological connection relationships match each other, the preliminary household-to-household relationship will be used as the household-to-household relationship of the low-voltage distribution network; if the topological connection relationships cannot match each other, conduct field verification to complete the identification of the household-to-household relationship.
[0097] The present invention also provides a computer medium, comprising a processor, a memory and a computer program for implementing the method of the present invention.
[0098] The present invention has the following beneficial effects:
[0099] The method for identifying the relationship between household transformers in a low-voltage distribution network of the present invention, by integrating the advantages of feature clustering and associated convolution methods, enables the method of the present invention to overcome the defect that the single voltage information is too similar to distinguish the belonging relationship, thereby optimizing the identification process and improving the identification accuracy, so that the method of the present invention can efficiently and accurately identify the relationship between household transformers in the distribution network; the method of the present invention proposes a topology correction method based on a knowledge graph, which supplements and improves the household transformer information of the low-voltage active distribution network through knowledge reasoning, reduces the errors and omissions of the household transformer relationship information, and facilitates the retrieval, viewing and analysis of the low-voltage distribution network data while improving the accuracy of topology identification. The method of the present invention can effectively reduce the time and space cost of household transformer relationship identification, improve the accuracy of identifying the complex topological structure of the low-voltage distribution network, and more accurately reflect the actual situation of the household transformer relationship in the low-voltage distribution network.
[0100] In addition to the above-described purposes, features and advantages, the present invention has other purposes, features and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0101] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:
[0102] Figure 1 It is a flow chart of a preferred embodiment of the present invention. DETAILED DESCRIPTION
[0103] The embodiments of the present invention are described in detail below with reference to the accompanying drawings, but the present invention can be implemented in many different ways as defined and covered by the claims.
[0104] See also Figure 1 In a preferred embodiment of the present invention, a method for identifying the relationship between household transformers in a low-voltage distribution network based on power balance and knowledge graph is provided, comprising the following steps:
[0105] S1, collecting first information and second information; the first information includes the location information and voltage information of the terminal user; the second information includes the distribution transformer data; and clustering the initial addresses of neighboring terminal users based on the first information. S1 specifically includes:
[0106] Collecting the first information and the second information includes:
[0107] Assume that there are m distribution transformers and n end users in the area to be identified. The set M of distribution transformers and the set N of end users can be expressed by the following relationship:
[0108] M = {1, 2, ..., m}
[0109] N = {1, 2, ..., n}
[0110] The location information and voltage information of the terminal users in the area to be identified are collected to obtain the first information; and the distribution transformer data are collected to obtain the second information.
[0111] The location information includes the user's address information in a preset format, and the preset format includes four gradually refined parts: cell information, building information, unit information, and room information; the voltage information includes the user's power information, active power information, and voltage information;
[0112] In a preferred embodiment of the present invention, clustering the initial addresses of neighboring terminal users according to the first information includes:
[0113] The terminal users whose location information, namely, the cell information, the building information, and the unit information, are the same in the first information are screened and clustered to obtain a user set U.
[0114] Calculate the voltage information of the end users in the user set U:
[0115] According to the room information in the user set U, the terminal users on the floor s are filtered and clustered to obtain the neighbor user set Us.
[0116] Use a central user node to replace the neighbor user set Us and get the power of the central user node Active Power And voltage data sequence Defining power and active power They are the total electricity and active power of all end users in the set Us:
[0117]
[0118] in, Indicates the power of the user in Us; Indicates the active power of the user in Us; and They respectively represent the electricity data and active power data of the central user node Us at the tth moment, t=1, 2, ..., W; T represents transposition.
[0119] The original voltage data of the terminal user in the first information is analyzed by the Z-Score method. To standardize:
[0120]
[0121] in, represents the standardized voltage value of the end user at time t; represents the mean voltage of the terminal users at time t; Represents the voltage standard deviation of the terminal user at time i.
[0122] Define the voltage of the central user node of Us is the mean voltage of all users in Us:
[0123]
[0124] Among them, k is the number of terminal users in Us; l∈US is the lth terminal user in Us.
[0125] Based on this, the voltage data sequence is obtained
[0126]
[0127] in, is the voltage data of the central user at the tth moment.
[0128] S2, clustering analysis is performed on the terminal user voltage time series data based on the Gaussian mixture model according to the initial address clustering and the second information, to obtain a preliminary classification result of the terminal user. S2 specifically includes:
[0129] S21. Initialize model parameters:
[0130] Define the number of clusters k cluster , k cluster The value is equal to the number of distribution transformers m; define the voltage data of the central user node of Us for The feature vector in W-dimensional sample space Gaussian mixture distribution Defined as:
[0131]
[0132] Among them, μ is the mean of the W-dimensional vector; θ is the W×W-dimensional covariance matrix; η jis the Gaussian mixing coefficient, which indicates the probability of selecting the jth mixing component, and its value is greater than 0; exp(·) is an exponential function.
[0133] By k cluster Gaussian mixture components, μ j and θ j is the mean vector and covariance matrix of the jth Gaussian component; at the beginning of clustering, the model parameters {(η j , μ j ,θ j )|1≤j≤k cluster}.
[0134] S22. Update model parameters:
[0135] In each iteration, the posterior probability Ψ of the sample belonging to each Gaussian component is calculated based on the current parameters. t，j , and then the model parameter η j , μ j ,θ j Perform iterative updates. S22 specifically includes:
[0136] Set the random variable g j ∈{1, 2, ..., k cluster} indicates generating samples Gaussian mixture components.
[0137] According to Bayes' theorem, we can get:
[0138]
[0139] Among them, t，j For sample The posterior probability generated by the jth Gaussian mixture component; μ r and θ r is the mean vector and covariance matrix of the rth Gaussian component; r is the random variable g j The rth element in .
[0140] To solve the model parameters {(η j , μ j ,θ j )|1≤j≤k cluster}, It can be further expressed as:
[0141]
[0142] The model parameters are iteratively optimized using the maximum expectation algorithm to achieve continuous updating. j , μ j ,θ j)|1≤j≤k cluster}Take the maximum value, From this we can get:
[0143]
[0144] Then we can get μ j :
[0145]
[0146] make Then we can find θ j :
[0147]
[0148] For the mixing coefficient η j , must satisfy η j ≥0, Its Lagrangian form is:
[0149]
[0150] By taking the derivative of the mixing coefficient and setting the inverse to 0, we can obtain η j :
[0151]
[0152] S23, according to Gaussian mixture distribution, Divide into k cluster Cluster sample The cluster The judgment formula can be expressed as:
[0153] γ i = argmaxΨ t，j , j∈{1,…,k cluster}
[0154]
[0155] That is, the preliminary division result of the user is obtained.
[0156] S3. Based on the preliminary division results and the power balance principle, an associated convolution optimization model is constructed and iteratively solved to obtain the preliminary household-to-household relationship in the low-voltage distribution system. S3 specifically includes:
[0157] Define the binary association variable x, then the connection relationship between the distribution transformer node j and the terminal user node i in the preliminary division result can be expressed as:
[0158]
[0159] Among them, Ω j is the set of users within the power supply range of low-voltage distribution transformer j and having electrical connectivity with distribution transformer j; i，j is the connection relationship between i and j. When end user i is within the power supply range of distribution transformer j and has an electrical connection with the distribution transformer, x j，i =1; On the contrary, when there is no electrical connection between the end user i and the distribution transformer j, x j , i =0.
[0160] In a low-voltage distribution transformer system including m distribution transformers and n end users, the sets of distribution transformers and end users are M = [1, 2, ..., j, ..., m] and N = [1, 2, ..., i, ..., n] respectively; for any low-voltage distribution transformer j∈M, define the association vector X j Describe the connectivity relationship between distribution transformer j and all users:
[0161] X j =[x j，1 , x j，2 , ..., x j，n ]
[0162] The association vector combination of m distribution transformers together constitutes a 0-1 adjacency matrix X that represents the connectivity relationship between all distribution transformers and end users in the low-voltage distribution system, that is, the connection relationship matrix:
[0163] X=[X 1 , X 2 , ..., X m ] T
[0164] There is a unique mapping relationship between distribution transformers and end users, then:
[0165]
[0166] Among them, x j，i Represents the element at row j and column i in X.
[0167] The associative convolution model describing the connection relationship between distribution transformer j and end users can be expressed as:
[0168]
[0169]
[0170] Among them, * represents the associated convolution operation. The convolution kernel slides on the input data matrix along the time axis with a sliding step of 1. The weighted sum of the coverage area is calculated each time, and the output data matrix is finally obtained; z N and p N Input matrix for electricity and power data; E M represents the power output matrix; p M represents the power output matrix; e N represents the power input matrix; p N represents the power input matrix; Indicates E M The element in row t of ; Indicates p M The element in row t of ; Indicates E M The jth element in the tth row; Indicates P M The jth element in the tth row; Indicates e N The i-th element in the t-th row; Indicates p N The ith element in the tth row.
[0171] The household-to-household relationship based on energy supply and demand balance, i.e. power balance, can be described as:
[0172] E M =e N *X+ξ
[0173] X T H=1 n
[0174] Where ξ is the error matrix, which is generated by the measurement, noise, and line loss during the energy transmission process of distribution transformer j; H is a matrix composed of m matrices D, H = [D, D, ..., D] T , D is the n×n dimensional identity matrix; 1 n is a 1×n dimensional matrix whose elements are all 1.
[0175] The household change relationship problem is transformed into an optimization problem, by solving the unknown binary variables x in the connection relationship matrix j，i Finally, the physical connection relationship between the distribution transformer and the end user is determined. In order to increase the solvability of the problem, the Softmax function is introduced to the binary variable x j，i Relax to get x′ j，i :
[0176]
[0177] X = [x′ j，i ] m×n
[0178]
[0179] Among them, e () It means to find the exponentiation with e as the base; h=1 is the starting number for the sum.
[0180] Considering the minimum sum of squares of the error between the energy supplied by the distribution transformer and the total energy consumption of the end user, the objective function is defined as:
[0181]
[0182] in, Represents the tth element of the power input matrix.
[0183] Solving minJ 1 , we get the matrix X, that is, we get the preliminary household-to-household relationship in the low-voltage distribution system.
[0184] S4. Based on the knowledge graph, the preliminary household-to-household relationship in the low-voltage distribution system is modified to obtain the household-to-household relationship of the low-voltage distribution network and complete the household-to-household relationship identification. S4 specifically includes:
[0185] S41. Obtain the location information of the terminal users in the area to be identified based on the knowledge graph, cluster the terminal users with the same cell information and building information, and obtain the terminal user set Ux of the Bth building.
[0186] S42. Count the number of distribution transformers N in the Bth building based on the knowledge graph B :
[0187]
[0188] S43, according to the capacity SN of the distribution transformer and the rated power factor of the distribution transformer The number of terminal users Nu of building B and the typical load PN of low-voltage users are used to obtain the number of transformers N in the building that meets the relevant national standards. D :
[0189]
[0190] S44. Solve for N B With N D The difference between B -N D ≥1, indicating that there is an error in the household change relationship of Building B, and preliminary confirmation requires on-site verification.
[0191] S45. After preliminary confirmation that field verification is required, use the elements in the knowledge graph to compare with the preliminary household-to-household relationship; if the topological connection relationships match each other, the preliminary household-to-household relationship will be used as the household-to-household relationship of the low-voltage distribution network; if the topological connection relationships cannot match each other, conduct field verification to complete the identification of the household-to-household relationship.
[0192] A preferred embodiment of the present invention further provides a computer medium, including a processor, a memory, and a computer program for implementing the method of the present invention.
[0193] In summary, the method for identifying the relationship between household transformers in a low-voltage distribution network of the present invention, by integrating the advantages of feature clustering and associated convolution methods, enables the method of the present invention to overcome the defect that the single voltage information is too similar to distinguish the belonging relationship, thereby optimizing the identification process and improving the identification accuracy, so that the method of the present invention can efficiently and accurately identify the relationship between household transformers in the distribution network; the method of the present invention proposes a topology correction method based on a knowledge graph, which supplements and improves the household transformer information of the low-voltage active distribution network through knowledge reasoning, reduces the errors and omissions of household transformer relationship information, and facilitates the retrieval, viewing and analysis of low-voltage distribution network data while improving the accuracy of topology identification. The method of the present invention can effectively reduce the time and space cost of household transformer relationship identification, improve the accuracy of identifying the complex topological structure of the low-voltage distribution network, and more accurately reflect the actual situation of the household transformer relationship in the low-voltage distribution network.
[0194] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A method for identifying household-to-substation relationships in a low-voltage distribution network based on power balance and knowledge graph, characterized in that: The following steps are involved: S1, collecting first information and second information; the first information includes the location information and voltage information of the terminal user; the second information includes the distribution transformer data; clustering the initial addresses of neighboring terminal users according to the first information; S2, performing cluster analysis on the voltage time series data of the terminal users based on the Gaussian mixture model according to the initial address clustering and the second information, and obtaining a preliminary classification result of the terminal users; S3. According to the preliminary division results and the power balance principle, an associated convolution optimization model is constructed, and an iterative solution is performed to obtain a preliminary household-to-household relationship in the low-voltage power distribution system; S4. Based on the knowledge graph, the preliminary household-to-household relationship in the low-voltage power distribution system is modified to obtain the household-to-household relationship of the low-voltage power distribution network, and complete the household-to-household relationship identification.
2. The method for identifying household-to-transformer relationships in a low-voltage distribution network based on power balance and knowledge graph according to claim 1 is characterized in that: The collecting of the first information and the second information in S1 includes: Assume that there are m distribution transformers and n end users in the area to be identified. The set M of distribution transformers and the set N of end users can be expressed by the following relationship: M={1,2,…,m} N={1,2,…,n} Collecting location information and voltage information of terminal users in the area to be identified to obtain first information; collecting distribution transformer data to obtain second information; The location information includes the user's address information in a preset format, and the preset format includes four gradually refined parts: cell information, building information, unit information and room information; the voltage information includes the user's power information, active power information and voltage information.
3. The method for identifying household-to-transformer relationships in a low-voltage distribution network based on power balance and knowledge graph according to claim 2 is characterized in that: The initial address clustering of neighboring terminal users according to the first information in S1 includes: Filter and cluster the terminal users whose location information, namely, cell information, building information, and unit information, in the first information, are the same, to obtain a user set U; Calculate the voltage information of the terminal users in the user set U: According to the room information in the user set U, terminal users on the floor s are filtered and clustered to obtain a neighbor user set Us; Use a central user node to replace the neighbor user set Us, and get the power of the central user node Active Power And voltage data sequence Defining power and active power They are the total electricity and active power of all end users in the set Us: in, Indicates the power of the user in Us; Indicates the active power of the user in Us; and They represent the electricity data and active power data of the central user node of Us at the tth moment, t=1,2,…,W; T represents the transposition; The original voltage data of the terminal user in the first information is analyzed by the Z-Score method. To standardize: in, represents the standardized voltage value of the end user at time t; represents the mean voltage of the terminal users at time t; It represents the voltage standard deviation of the terminal user at time i; Define the voltage of the central user node of Us is the mean voltage of all users in Us: Where k is the number of end users in Us; l∈US is the lth end user in Us; The voltage data sequence is obtained accordingly in, is the voltage data of the central user at the tth moment.
4. The method for identifying household-to-transformer relationships in a low-voltage distribution network based on power balance and knowledge graph according to claim 3 is characterized in that: The S2 includes: S21. Initialize model parameters: Define the number of clusters k cluster , k cluster The value is equal to the number of distribution transformers m; define the voltage data of the central user node of Us for The feature vector in W-dimensional sample space Gaussian mixture distribution Defined as: Among them, μ is the mean of the W-dimensional vector; θ is the W×W-dimensional covariance matrix; η j is the Gaussian mixing coefficient, which indicates the probability of selecting the jth mixing component, and its value is greater than 0; exp(·) is an exponential function; By k cluster Gaussian mixture components, μ j and θ j is the mean vector and covariance matrix of the jth Gaussian component; at the beginning of clustering, the model parameters {(η j , μ j ,θ j )|1≤j≤k cluster }; S22. Update model parameters: In each iteration, the posterior probability Ψ of the sample belonging to each Gaussian component is calculated based on the current parameters. t，j , and then the model parameter η j , μ j ,θ j Perform iterative updates; S23. According to Gaussian mixture distribution, Divide into k cluster Cluster sample The cluster The judgment formula can be expressed as: c i =arg max Ψ t，j ,j∈{1,…,k cluster } That is, the preliminary division result of the user is obtained.
5. The method for identifying household-to-transformer relationships in a low-voltage distribution network based on power balance and knowledge graph according to claim 4 is characterized in that: The S22 includes: Set the random variable g j ∈{1, 2, ..., k cluster } indicates generating samples Gaussian mixture components; According to Bayes' theorem, we can get: Among them, t，j For sample The posterior probability generated by the jth Gaussian mixture component; μ r and θ r is the mean vector and covariance matrix of the rth Gaussian component; r is the random variable g j The rth element in ; To solve the model parameters {(η j , μ j ,θ j )|1≤j≤k cluster }, It can be further expressed as: The model parameters are iteratively optimized using the maximum expectation algorithm to achieve continuous updating. j , μ j ,θ j )|1≤j≤k cluster }Take the maximum value, From this we can get: Then we can get μ j : make Then we can find θ j : For the mixing coefficient η j , must meet Its Lagrangian form is: By taking the derivative of the mixing coefficient and setting the inverse to 0, we can obtain η j :
6. The method for identifying household-to-transformer relationships in a low-voltage distribution network based on power balance and knowledge graph according to claim 5 is characterized in that: The S3 includes: Define a binary association variable x, then the connection relationship between the distribution transformer node j and the terminal user node i in the preliminary division result can be expressed as: Among them, Ω j is the set of users within the power supply range of low-voltage distribution transformer j and having electrical connectivity with distribution transformer j; i，j is the connection relationship between i and j. When end user i is within the power supply range of distribution transformer j and has an electrical connection with the distribution transformer, x j，i =1; On the contrary, when there is no electrical connection between the end user i and the distribution transformer j, x j，i =0; In a low-voltage distribution transformer system including m distribution transformers and n end users, the sets of distribution transformers and end users are M = [1, 2, ..., j, ..., m] and N = [1, 2, ..., i, ..., n] respectively; for any low-voltage distribution transformer j∈M, define the association vector X j Describe the connectivity relationship between distribution transformer j and all users: X j =[x j，1 ,x j，2 ,...,x j，n ] The association vector combination of m distribution transformers together constitutes a 0-1 adjacency matrix X that represents the connectivity relationship between all distribution transformers and end users in the low-voltage distribution system, that is, the connection relationship matrix: X=[X1,X2,...,X m ] T There is a unique mapping relationship between distribution transformers and end users, then: Among them, x j，i represents the element in the jth row and ith column of X; The associative convolution model describing the connection relationship between distribution transformer j and end users can be expressed as: Among them, * represents the associated convolution operation. The convolution kernel slides on the input data matrix along the time axis with a sliding step of 1. The weighted sum of the coverage area is calculated each time, and the output data matrix is finally obtained; z N and p N Input matrix for electricity and power data; E M Represents the power output matrix; P M represents the power output matrix; e N represents the power input matrix; p N represents the power input matrix; Indicates E M The element in row t of ; Indicates P M The element in row t of ; Indicates E M The jth element in the tth row; Indicates P M The jth element in the tth row; Indicates e N The i-th element in the t-th row; Indicates p N The i-th element in the t-th row; The household-to-household relationship based on energy supply and demand balance, i.e. power balance, can be described as: AND M =and N *X+ξ X T H=1 n Where ξ is the error matrix, which is generated by the measurement, noise, and line loss during the energy transmission process of distribution transformer j; H is a matrix composed of m matrices D, H = [D, D, ..., D] T , D is the n×n dimensional identity matrix; 1 n is a 1×n dimensional matrix whose elements are all 1; The household change relationship problem is transformed into an optimization problem, by solving the unknown binary variables x in the connection relationship matrix j，i Finally, the physical connection relationship between the distribution transformer and the end user is determined. In order to increase the solvability of the problem, the Softmax function is introduced to the binary variable x j，i Relax to get x′ j，i : X=[x′ j，i ] m×n Among them, e () It means to calculate the power with e as the base; h=1 is the starting number for summation; Considering the minimum sum of squares of the error between the energy supplied by the distribution transformer and the total energy consumption of the end user, the objective function is defined as: in, Represents the tth element of the power input matrix; Solve for min J1 and get the matrix X, that is, the preliminary household-to-household relationship in the low-voltage distribution system.
7. The method for identifying household-to-transformer relationships in a low-voltage distribution network based on power balance and knowledge graph according to claim 6 is characterized in that: The S4 includes: S41, obtaining the location information of the terminal users in the to-be-identified area according to the knowledge graph, clustering the terminal users with the same cell information and building information, and obtaining the terminal user set Ux of the Bth building; S42. Count the number of distribution transformers N in the Bth building based on the knowledge graph B : S43, according to the capacity of the distribution transformer S N , Rated power factor of distribution transformer The number of terminal users Nu and the typical load P of low-voltage users in Building B N , get the number of transformers N in the building that meets the relevant national standards D : S44. Solve for N B With N D The difference between B -N D ≥1, indicating that there is an error in the household change relationship of Building B, and preliminary confirmation requires on-site verification; S45. After preliminary confirmation that field verification is required, use the elements in the knowledge graph to compare with the preliminary household-to-household relationship; if the topological connection relationships match each other, use the preliminary household-to-household relationship as the household-to-household relationship of the low-voltage distribution network; if the topological connection relationships cannot match each other, perform field verification to complete the household-to-household relationship identification.
8. A computer medium, characterized in that The invention comprises a processor, a memory and a computer program for implementing the method according to any one of claims 1 to 7.