Low-voltage transformer area topology identification method and apparatus, storage medium, and chip device

Abstract

Disclosed in the present invention are a low-voltage transformer area topology identification method and apparatus, a storage medium, and a chip device. The low-voltage transformer area topology identification method comprises: obtaining electricity amount time series data of devices to be identified in a low-voltage transformer area to obtain original electricity amount data; performing dimension reduction processing on the original electricity amount data to obtain low-dimensional electricity amount data of branch boxes and low-dimensional electricity amount data of meter boxes; determining a transformer area master meter on the basis of the low-dimensional electricity amount data of the branch boxes, updating branch boxes waiting for determination of a connection relationship to obtain a first connection relationship of the branch boxes, and obtaining final-stage branch boxes; on the basis of low-dimensional electricity amount data of the final-stage branch boxes and the low-dimensional electricity amount data of the meter boxes, determining a second connection relationship of the meter boxes; and on the basis of the first and second connection relationships, obtaining a topological structure of the low-voltage transformer area. According to the low-voltage transformer area topology identification method in embodiments of the present invention, the interference caused by redundant information in high-dimensional data can be reduced, and the topological structure of the low-voltage transformer area can be accurately obtained.

Description

Low-voltage station area topology identification method, device, storage medium and chip device Technical Field

[0001] The present invention relates to the technical field of power distribution networks, and in particular to a low-voltage substation topology identification method, a low-voltage substation topology identification device, a non-temporary computer-readable storage medium, and a chip device. Background Art

[0002] The physical topology of a low-voltage substation accurately reflects the connections between distribution transformers, lines, and users within the substation. As shown in Figure 1, the low-voltage substation topology consists of the substation-transformer layer, the branch layer, and the meter box layer. The substation-transformer layer includes a substation master meter, the branch layer includes multiple branch boxes, and the meter box layer includes multiple meter boxes. Real-time understanding of the substation master meter-branch box-meter box topology provides a crucial foundation for lean power system management, line fault location, and substation line loss analysis.

[0003] However, low-voltage substations have long had the problem of chaotic topological relationships, and the topological relationships need to be clarified to support advanced power applications. At present, the relevant topology identification technologies mainly include the following three: 1. Pulse current method, which performs topological identification by injecting characteristic signals into the substation. This method wastes manpower and material resources, and the injection of power frequency signals will cause hidden dangers to the power grid safety. 2. Carrier communication method, which determines the correspondence between the nodes of the circuit by sending and receiving power line carrier signals, but due to the influence of power load, the signal credibility is low and the topological structure cannot be accurately identified. 3. Data mining method, which determines the connection relationship between substation equipment based on the similarity between power consumption information such as voltage, power, and electricity, but the recognition accuracy of the data mining method in related technologies is not high.

[0004] Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems in the related art to a certain extent. To this end, the present invention aims to provide a low-voltage area topology identification method, apparatus, storage medium and chip device to accurately identify the topology structure of the low-voltage area.

[0006] In order to achieve the above-mentioned purpose, the first embodiment of the present invention proposes a low-voltage substation topology identification method, the method comprising: obtaining the time series data of the power of each device to be identified in the low-voltage substation to obtain the original power data; performing dimensionality reduction processing on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box; determining the substation total table according to the low-dimensional power data of the branch box, updating the branch box with the connection relationship to be determined, and grouping the branch boxes with the connection relationship to be determined to obtain multiple combinations, and according to the low-dimensional power data of the branch box in each combination and the target branch box The low-dimensional electricity data is used to determine the target combination as the next layer of the target branch box, and return to the step of updating the branch box with the connection relationship to be determined, until the last layer of the low-voltage substation branch box is reached, and the first connection relationship of the branch box is obtained, and the last-level branch box is obtained, wherein the target branch box is the upper-level branch box of the branch box with the connection relationship to be determined; according to the low-dimensional electricity data of the last-level branch box and the low-dimensional electricity data of the meter box, the second connection relationship of the meter box is determined; according to the first connection relationship and the second connection relationship, the topological structure of the low-voltage substation is obtained.

[0007] In addition, the low-voltage area topology identification method according to the embodiment of the present invention may also have the following additional technical features:

[0008] According to one embodiment of the present invention, determining the substation summary table based on the low-dimensional electricity data of the branch box includes: calculating the average value of the electricity data of each branch box based on the low-dimensional electricity data of the branch box, and taking the branch box corresponding to the maximum average value as the substation summary table.

[0009] According to one embodiment of the present invention, the grouping of the branch boxes whose connection relationships are to be determined to obtain multiple combinations includes: randomly selecting 1, ..., e-1, e branch boxes from the branch boxes whose connection relationships are to be determined to group them, and obtaining combinations, where K is the number of branch boxes whose connection relationship is to be determined.

[0010] According to one embodiment of the present invention, the low-dimensional electricity data is time series data, and the target combination is determined based on the low-dimensional electricity data of the branch boxes in each combination and the low-dimensional electricity data of the target branch box, including: for each combination, adding the electricity data at the same time in the low-dimensional electricity data of each branch box in the combination to obtain the cumulative electricity data of the combination; calculating the Pearson correlation coefficient of the cumulative electricity data of each combination and the low-dimensional electricity data of the target branch box respectively; and taking the combination corresponding to the maximum Pearson correlation coefficient as the target combination.

[0011] According to one embodiment of the present invention, after obtaining the first connection relationship of the branch box, the method further includes: determining whether there is a leaky connected branch box in the first connection relationship; if so, respectively calculating the Pearson correlation coefficient of the low-dimensional power data of the leaky connected branch box and the cumulative power sequence corresponding to each target combination; determining the position of the leaky connected branch box in the first connection relationship according to the target combination corresponding to the maximum Pearson correlation coefficient, and updating the first connection relationship according to the position.

[0012] According to one embodiment of the present invention, the method also includes: determining whether there is a branch box with a direct connection relationship in the first connection relationship; if so, sending a high-frequency signal to the low-voltage substation through the power carrier, and obtaining the signal strength of the branch box with the direct connection relationship; determining the superior-subordinate relationship of the branch box with the direct connection relationship based on the signal strength, and updating the first connection relationship based on the superior-subordinate relationship.

[0013] According to one embodiment of the present invention, the second connection relationship of the meter boxes is determined based on the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, including: for each of the final branch boxes, based on the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, constructing a linear relationship between the final branch box and all the meter boxes, converting the linear relationship into a convex optimization model, and solving the convex optimization model to obtain the meter box connected to the final branch box; according to the meter boxes connected to each of the final branch boxes, the second connection relationship of the meter boxes is obtained.

[0014] According to one embodiment of the present invention, converting the linear relationship into a convex optimization model includes: converting the linear relationship into an optimization model with the smallest solution error; replacing the L0 norm in the optimization model with the L2 norm to obtain the convex optimization model.

[0015] According to one embodiment of the present invention, the linear relationship is:

[0016] The optimization model is:

[0017] st x m,i ∈{0,1}

[0018] The convex optimization model is:

[0019] st x m,i ∈[0,1]

[0020] in, represents the low-dimensional power data of the i-th last-stage branch box, T' represents the time number of the low-dimensional power data, represents the low-dimensional power data of M meter boxes, x=[x 1,i ,x 2,i ,…,x m,i ,…,x M,i ] T It represents the meter box that builds a linear relationship with the i-th said last-stage branch box. When the m-th said meter box and the i-th said last-stage branch box have a connection relationship, x m,i =1, otherwise x m,i =0, β represents the error caused by random noise interference in the power measurement value. represents the L0 norm, represents the L2 norm.

[0021] According to one embodiment of the present invention, the dimensionality reduction processing of the original electric power data to obtain low-dimensional electric power data of the branch box and low-dimensional electric power data of the meter box includes: calculating the Gaussian probability distribution matrix of the original electric power data; constructing low-dimensional data, initializing the low-dimensional data, and calculating the t-distribution matrix of the low-dimensional data, wherein the number of moments of the low-dimensional data is less than the number of moments of the original electric power data; calculating the KL divergence of the Gaussian probability distribution matrix and the t-distribution matrix as a loss function; and solving the low-dimensional data using the gradient descent method based on the loss function to obtain the low-dimensional electric power data of the branch box and low-dimensional electric power data of the meter box.

[0022] According to one embodiment of the present invention, the calculating the Gaussian probability distribution matrix of the original electric power data includes: calculating the similarity conditional probability between each pair of data in the original electric power data, and calculating a first joint probability based on the similarity conditional probability, and obtaining the Gaussian probability distribution matrix of the original electric power data based on the first joint probability; the calculating the t-distribution matrix of the low-dimensional data includes: calculating the second joint probability of the low-dimensional data through the t-distribution function, wherein the initial value of the low-dimensional data is obtained by random initialization; and obtaining the t-distribution matrix of the low-dimensional data based on the second joint probability.

[0023] According to one embodiment of the present invention, the low-dimensional data is solved by the gradient descent method based on the loss function, including: according to the loss function, the low-dimensional data is calculated by the gradient descent method to obtain a gradient function; according to the gradient function, a gradient vector is obtained, and the low-dimensional data is updated according to the gradient vector; it is determined whether the updated low-dimensional data meets the iteration end condition; if not, returning to the step of calculating the t distribution matrix of the low-dimensional data; if so, using the current low-dimensional data as the low-dimensional electrical quantity data.

[0024] According to one embodiment of the present invention, the similarity conditional probability is calculated by the following formula:

[0025] in, represents the raw power data, represents the time series data of the power of the i-th device to be identified, D represents the total number of the devices to be identified, P j|i represents the similarity conditional probability of the i-th device to be identified being distributed around the j-th device to be identified, σ i Indicates The variance of a normal distribution centered at .

[0026] The first joint probability is calculated by the following formula:

[0027] Among them, p ij It represents the probability similarity between the i-th device to be identified and the j-th device to be identified.

[0028] The Gaussian probability distribution matrix is ​​expressed as: P = [p1, p2, ... p i ,…,p D ], where p i =[p i1 ,p i2 ,…,p iw ,…,p iD ], w≠i.

[0029] The second joint probability is calculated by the following formula:

[0030] Where Q = [q1;q2;…q i ;…;q D ]∈R D×T′ represents the low-dimensional data, q i ∈R 1×T′ represents the low-dimensional data of the i-th device to be identified, and T' represents the time number of the low-dimensional data.

[0031] The t distribution matrix is ​​expressed as: R = [r1, r2, ..., r i ,…,r D ], where r i =[r i1 ,r i2 ,…r iv ,…,r iD ], v≠i.

[0032] According to one embodiment of the present invention, the loss function is calculated by the following formula:

[0033] Among them, C i Represents the loss function of the i-th device to be identified.

[0034] The gradient function is calculated by the following formula:

[0035] The gradient vector is expressed as:

[0036] According to one embodiment of the present invention, iterative calculation is performed using the following formula:

[0037] Among them, q t represents the result of the t-th iteration of low-dimensional electrical data, η represents the learning rate, and α represents the momentum factor.

[0038] To achieve the above-mentioned purpose, the second embodiment of the present invention proposes a low-voltage substation topology identification device, characterized in that the device includes: a first acquisition module, used to obtain the power time series data of each device to be identified in the low-voltage substation, and obtain the original power data; a data processing module, used to perform dimensionality reduction processing on the original power data, and obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box; a first determination module, used to determine the substation total table according to the low-dimensional power data of the branch box, update the branch box with the connection relationship to be determined, and group the branch boxes with the connection relationship to be determined to obtain multiple combinations, and according to the low-dimensional power of the branch box in each combination, The target combination is determined based on the quantity data and the low-dimensional quantity data of the target branch box, and the branch box with the to-be-determined connection relationship is re-updated as the next layer of the target branch box, until the last layer of the low-voltage substation branch box is reached, the first connection relationship of the branch box is obtained, and the last-level branch box is obtained, wherein the target branch box is the upper-level branch box of the branch box with the to-be-determined connection relationship; the second determination module is used to determine the second connection relationship of the meter box according to the low-dimensional quantity data of the last-level branch box and the low-dimensional quantity data of the meter box; the third determination module is used to obtain the topological structure of the low-voltage substation according to the first connection relationship and the second connection relationship.

[0039] In addition, the low-voltage substation topology identification device according to the embodiment of the present invention may also have the following additional technical features:

[0040] According to one embodiment of the present invention, when determining the substation total table based on the low-dimensional electricity data of the branch box, the first determination module is used to: calculate the average value of the electricity data of each branch box based on the low-dimensional electricity data of the branch box, and use the branch box corresponding to the maximum average value as the substation total table.

[0041] According to one embodiment of the present invention, when the first determining module groups the branch boxes whose connection relationships are to be determined to obtain multiple combinations, it is configured to: randomly select 1, ..., e-1, e branch boxes from the branch boxes whose connection relationships are to be determined to group them, and obtain combinations, where K is the number of branch boxes whose connection relationship is to be determined.

[0042] According to one embodiment of the present invention, the low-dimensional electric power data is time series data. When the first determination module determines the target combination based on the low-dimensional electric power data of the branch boxes in each of the combinations and the low-dimensional electric power data of the target branch box, it is used to: for each of the combinations, add the electric power data at the same time in the low-dimensional electric power data of each branch box in the combination to obtain the cumulative electric power data of the combination; calculate the Pearson correlation coefficient of the cumulative electric power data of each of the combinations and the low-dimensional electric power data of the target branch box respectively; and take the combination corresponding to the maximum Pearson correlation coefficient as the target combination.

[0043] According to one embodiment of the present invention, the device also includes: a first judgment module, used to determine whether there is a leaky connection branch box in the first connection relationship; a first calculation module, used to calculate the Pearson correlation coefficient of the low-dimensional power data of the leaky connection branch box and the cumulative power sequence corresponding to each target combination when there is a leaky connection branch box; a fourth determination module, used to determine the position of the leaky connection branch box in the first connection relationship according to the target combination corresponding to the maximum Pearson correlation coefficient, and update the first connection relationship according to the position.

[0044] According to one embodiment of the present invention, the device also includes: a second judgment module, used to judge whether there is a branch box with a direct connection relationship in the first connection relationship; a second acquisition module, used to send a high-frequency signal to the low-voltage substation through a power carrier when there is a branch box with a direct connection relationship, and obtain the signal strength of the branch box with a direct connection relationship; a fifth determination module, used to determine the superior-subordinate relationship of the branch box with a direct connection relationship based on the signal strength, and update the first connection relationship based on the superior-subordinate relationship.

[0045] According to one embodiment of the present invention, when the second determination module determines the second connection relationship of the meter box based on the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, it is used to: for each of the final branch boxes, construct a linear relationship between the final branch box and all the meter boxes based on the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, and convert the linear relationship into a convex optimization model, and solve the convex optimization model to obtain the meter box connected to the final branch box; according to the meter boxes connected to each of the final branch boxes, obtain the second connection relationship of the meter boxes.

[0046] According to one embodiment of the present invention, when converting the linear relationship into a convex optimization model, the second determination module is used to: convert the linear relationship into an optimization model with the smallest solution error; and replace the L0 norm in the optimization model with the L2 norm to obtain the convex optimization model.

[0047] According to one embodiment of the present invention, the linear relationship is:

[0048] The optimization model is:

[0049] st x m,i ∈{0,1}

[0050] The convex optimization model is:

[0051] st x m,i ∈[0,1]

[0052] in, represents the low-dimensional power data of the i-th last-stage branch box, T' represents the time number of the low-dimensional power data, represents the low-dimensional power data of M meter boxes, x=[x 1,i ,x 2,i ,…,x m,i ,…,x M,i ] T It represents the meter box that builds a linear relationship with the i-th said last-stage branch box. When the m-th said meter box and the i-th said last-stage branch box have a connection relationship, x m,i =1, otherwise x m,i =0, β represents the error caused by random noise interference in the power measurement value. represents the L0 norm, represents the L2 norm.

[0053] According to one embodiment of the present invention, when the data processing module performs dimensionality reduction processing on the original electric power data to obtain low-dimensional electric power data of the branch box and low-dimensional electric power data of the meter box, it is used to: calculate the Gaussian probability distribution matrix of the original electric power data; construct low-dimensional data, initialize the low-dimensional data, and calculate the t-distribution matrix of the low-dimensional data, wherein the number of moments of the low-dimensional data is less than the number of moments of the original electric power data; calculate the KL divergence of the Gaussian probability distribution matrix and the t-distribution matrix as a loss function; and solve the low-dimensional data using the gradient descent method based on the loss function to obtain the low-dimensional electric power data of the branch box and the low-dimensional electric power data of the meter box.

[0054] According to one embodiment of the present invention, when calculating the Gaussian probability distribution matrix of the original electric power data, the data processing module is used to: calculate the similarity conditional probability between each pair of data in the original electric power data, and calculate the first joint probability based on the similarity conditional probability, and obtain the Gaussian probability distribution matrix of the original electric power data based on the first joint probability; when calculating the t-distribution matrix of the low-dimensional data, the data processing module is used to: calculate the second joint probability of the low-dimensional data through the t-distribution function, wherein the initial value of the low-dimensional data is obtained by random initialization; and obtain the t-distribution matrix of the low-dimensional data based on the second joint probability.

[0055] According to one embodiment of the present invention, when the data processing module uses the gradient descent method to solve the low-dimensional data based on the loss function, it is used to: calculate the low-dimensional data by the gradient descent method according to the loss function to obtain a gradient function; obtain a gradient vector according to the gradient function, and update the low-dimensional data according to the gradient vector; determine whether the updated low-dimensional data meets the iteration end condition; if not, return to the step of calculating the t distribution matrix of the low-dimensional data; if so, use the current low-dimensional data as the low-dimensional electrical quantity data.

[0056] According to one embodiment of the present invention, the data processing module calculates the similarity conditional probability using the following formula:

[0057] in, represents the raw power data, represents the time series data of the power of the i-th device to be identified, D represents the total number of the devices to be identified, P j|i represents the similarity conditional probability of the i-th device to be identified being distributed around the j-th device to be identified, σ i Indicates The variance of a normal distribution centered at .

[0058] The data processing module calculates the first joint probability using the following formula:

[0059] Among them, p ij It represents the probability similarity between the i-th device to be identified and the j-th device to be identified.

[0060] The Gaussian probability distribution matrix is ​​expressed as: P = [p1, p2, ... p i ,…,p D ], where p i =[p i1 ,p i2 ,…,p iw ,…,p iD ], w≠i.

[0061] The data processing module calculates the second joint probability using the following formula:

[0062] Where Q = [q1;q2;…q i ;…;q D ]∈R D×T′ represents the low-dimensional data, q i ∈R 1×T′ represents the low-dimensional data of the i-th device to be identified, and T' represents the time number of the low-dimensional data.

[0063] The t distribution matrix is ​​expressed as: R = [r1, r2, ..., r i ,…,r D ], where r i =[r i1 ,r i2 ,…r iv ,…,r iD ], v≠i.

[0064] According to one embodiment of the present invention, the data processing module calculates the loss function using the following formula:

[0065] Among them, C i Represents the loss function of the i-th device to be identified.

[0066] The data processing module calculates the gradient function using the following formula:

[0067] The gradient vector is expressed as:

[0068] According to one embodiment of the present invention, the data processing module performs iterative calculations using the following formula:

[0069] Among them, q t represents the result of the t-th iteration of low-dimensional electrical data, η represents the learning rate, and α represents the momentum factor.

[0070] To achieve the above-mentioned purpose, a non-temporary computer-readable storage medium is proposed in an embodiment of the third aspect of the present invention, on which a computer program is stored. When the program is executed by a processor, the above-mentioned low-voltage substation topology identification method is implemented.

[0071] To achieve the above-mentioned purpose, the fourth embodiment of the present invention proposes a chip device, including a memory, a processor and a computer program stored in the memory. When the computer program is executed by the processor, the above-mentioned low-voltage area topology identification method is implemented.

[0072] The beneficial effects of the low-voltage substation topology identification method, device, storage medium and chip device of the embodiment of the present invention are: 1. The use of electricity data as a numerical condition for judging the connection relationship of each target to be identified in the substation overcomes the influence of the high time synchronization requirements of the system when using instantaneous data such as current and power, and is easy to implement. 2. By performing dimensionality reduction processing on the original electricity data, the interference caused by redundant information in the high-dimensional data can be reduced, and the first connection relationship and the second connection relationship are obtained respectively through the low-dimensional electricity data, so that the topological structure of the low-voltage substation can be accurately obtained; 3. Different topology identification methods are adopted for the substation branch layer and the meter box layer, respectively, to ensure the effectiveness and reliability of the method; 4. When judging the first connection relationship, for the branch box that is missed, its location is determined according to the existing target combination. Furthermore, for the branch box with a direct connection relationship, the power line carrier signal attenuation characteristics are used to determine the superior and subordinate relationship between the two, which generally improves the overall identification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] FIG1 is a schematic diagram of a low-voltage station topology structure in related art;

[0074] FIG2 is a schematic flow chart of a method for identifying a low-voltage area topology according to an embodiment of the present invention;

[0075] 3 is a schematic diagram of a process for determining a target combination based on low-dimensional power data of branch boxes in each combination and low-dimensional power data of a target branch box according to an embodiment of the present invention;

[0076] 4 is a schematic diagram of a flow chart of determining a second connection relationship of a meter box based on low-dimensional power data of a final branch box and low-dimensional power data of a meter box according to an embodiment of the present invention;

[0077] FIG5 is a schematic diagram of a process for performing dimensionality reduction processing on raw power data according to an embodiment of the present invention;

[0078] 6 is a schematic structural diagram of a low-voltage area topology identification device according to an embodiment of the present invention;

[0079] 7 is a schematic diagram of the application process of the low-voltage area topology identification system according to an embodiment of the present invention;

[0080] FIG8 is a schematic structural diagram of a chip device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0081] The following describes embodiments of the present invention in detail, examples of which 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 intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0082] The following describes a low-voltage platform area topology identification method, apparatus, storage medium, and chip device according to embodiments of the present invention with reference to the accompanying drawings.

[0083] FIG2 is a flow chart of a method for identifying a low-voltage substation topology according to an embodiment of the present invention.

[0084] As shown in Figure 2, the low-voltage area topology identification method includes:

[0085] S1, obtain the time series data of the power of each device to be identified in the low-voltage area to obtain the original power data.

[0086] The raw power data may contain noise information.

[0087] S2, performing dimensionality reduction processing on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box.

[0088] As an example, the dimensionality reduction processing method may be principal component analysis or linear discriminant analysis.

[0089] It should be noted that dimensionality reduction processing can reduce the dimension of data, reduce data redundancy, and weaken the impact of noise.

[0090] S3, determine the substation total table based on the low-dimensional power data of the branch box, update the branch box with the connection relationship to be determined, and group the branch boxes with the connection relationship to be determined to obtain multiple combinations, and determine the target combination based on the low-dimensional power data of the branch boxes in each combination and the low-dimensional power data of the target branch box, as the next layer of the target branch box, return to the step of updating the branch box with the connection relationship to be determined, until the last layer of the low-voltage substation branch box is reached, the first connection relationship of the branch box is obtained, and the last-level branch box is obtained, wherein the target branch box is the branch box of the upper layer of the branch box with the connection relationship to be determined.

[0091] Specifically, the first connection relationship of the branch boxes can be determined based on the low-dimensional power data of the branch boxes and the principle of conservation of electrical energy. The principle of conservation of electrical energy states that within any time interval, the power of the measuring switch of the outgoing cabinet of a branch box and the measuring switch of its incoming cabinet are approximately equal at the same time.

[0092] More specifically, when determining the branch boxes in the first layer of the branch layer, all branch boxes except the substation master meter are considered as branch boxes whose connection relationships are to be determined. Subsequently, when determining the branch boxes in other layers of the branch layer, all branch boxes except the branch boxes whose connection relationships have been determined are considered as branch boxes whose connection relationships are to be determined.

[0093] More specifically, the branch boxes whose connection relationship is to be determined are grouped to obtain multiple combinations, including: randomly selecting 1, ..., e-1, e branch boxes from the branch boxes whose connection relationship is to be determined to group them, and obtaining combinations, where K is the number of branch boxes whose connection relationships are to be determined.

[0094] As an example, first, among the branch boxes after removing the total meter of the substation, randomly select 1, ..., e-1, e branch boxes to form a group, and get combinations, where N is the total number of branch boxes; then, among the branch boxes after removing the first layer of branch boxes in the area master table and the branch layer, randomly select 1, ..., z-1, z branch boxes to form a group, and get combinations, where y is the number of branch boxes in the first layer of the branch layer.

[0095] As an example, first determine the substation total meter, take the substation total meter as the target branch box, and determine the meter box on the first layer in the branch layer; then, take the meter box on the first layer in the branch layer as the target branch box in turn, determine the meter box on the first layer in the branch layer and its corresponding relationship, and cycle in sequence until reaching the last layer of the low-voltage substation branch layer.

[0096] S4, determining a second connection relationship between the meter boxes according to the low-dimensional power data of the last-stage branch box and the low-dimensional power data of the meter box.

[0097] S5. Obtain a topological structure of the low-voltage area according to the first connection relationship and the second connection relationship.

[0098] The low-voltage substation topology identification method of an embodiment of the present invention can reduce the interference caused by redundant information in high-dimensional data by performing dimensionality reduction processing on the original power data, and obtain the first connection relationship and the second connection relationship respectively through the low-dimensional power data, so as to accurately obtain the topological structure of the low-voltage substation, thereby generally improving the overall identification accuracy.

[0099] In some embodiments of the present invention, the substation summary table is determined based on the low-dimensional electricity data of the branch box, including: calculating the average value of the electricity data of each branch box based on the low-dimensional electricity data of the branch box, and taking the branch box corresponding to the maximum average value as the substation summary table.

[0100] In this embodiment, the average value of the power data of each branch box is calculated, and the data is sorted to use the branch box corresponding to the maximum average value as the substation total table, so that the substation total table in the branch box can be quickly determined.

[0101] In some embodiments of the present invention, the low-dimensional power data is time series data. As shown in FIG3 , the target combination is determined based on the low-dimensional power data of the branch boxes in each combination and the low-dimensional power data of the target branch box, including:

[0102] S341 , for each combination, adding the power data at the same time in the low-dimensional power data of each branch box in the combination to obtain the accumulated power data of the combination.

[0103] It should be noted that It means that each of the K groups has only one branch box, and the low-dimensional power data of the branch box is used as the cumulative power data of the combination.

[0104] S342 , respectively calculating the Pearson correlation coefficient between the accumulated power data of each combination and the low-dimensional power data of the target branch box.

[0105] Specifically, the Pearson correlation coefficient is calculated using the following formula:

[0106] in, represents the charge sequence of the mth branch box or target combination, cov(ε,τ) represents the covariance of ε and τ, and the symbol D(h) represents the variance of h.

[0107] S343, taking the combination corresponding to the maximum Pearson correlation coefficient as the target combination.

[0108] As an example, assume that the number of devices to be identified is D, where N represents the number of branch boxes and M represents the number of meter boxes. The low-dimensional power data Q at the branch layer is B Expressed as:

[0109] in, is the low-dimensional power data of the i-th branch box.

[0110] The power sequence of each branch box Take the average value and record it as B1, B2,…, B N, and sort them from large to small according to the mean value, take the branch box corresponding to the maximum mean value as the substation area summary table of the substation layer, and then judge the connection relationship of the branch boxes at each level in the branch layer in turn.

[0111] In this embodiment, the Pearson correlation coefficient ρ (-1≤ρ≤1) is used to measure the similarity between the target combination and the target branch box. The larger the ρ, the more likely the target combination and the target branch box are parent-child nodes. Therefore, the connection relationship between each branch box in the branch layer can be determined in sequence through the above method.

[0112] Taking Figure 1 as an example, the power consumption of branch box B31 is approximately equal to the sum of B41, B42, and B43. Therefore, the child nodes found by branch box B21 may be B31, B32, and B33, or B41, B42, and B43 (because B31 = B41 + B42 + B43), and B32 and B33. If B21 determines that its child nodes are the latter, B31 will be missed during the topology identification process.

[0113] In some embodiments of the present invention, after obtaining the first connection relationship of the branch boxes, the low-voltage substation topology identification method further includes:

[0114] S311: Determine whether there is a missing branch box in the first connection relationship.

[0115] S312: If it exists, calculate the Pearson correlation coefficient between the low-dimensional power data of the missing branch box and the cumulative power sequence corresponding to each target combination.

[0116] S313 , determining the position of the missing-connection branch box in the first connection relationship according to the target combination corresponding to the maximum Pearson correlation coefficient, and updating the first connection relationship according to the position.

[0117] The missing connection branch box may be connected between the target combination and the target branch box previously corresponding to the target combination.

[0118] In this embodiment, by acquiring all the leaked branch boxes, re-determining the connection relationship of the leaked branch boxes, and updating the connection relationship to the first connection relationship, the integrity of the substation topology structure can be ensured.

[0119] Taking Figure 1 as an example, there's only a slight difference in the power levels between branch boxes B36 and B4x. When determining the next-level branch box, branch box B22 cannot accurately determine whether its subordinate nodes are (B34, B35, B36) or (B34, B35, B4x). If it determines (B34, B35, B4x), B36 will be missed during topology identification. Using the above method, we can determine that the branch box with the highest correlation with B36 is B4x. Since there's only one branch box with the highest correlation with B36, we can conclude that B36 and B4x are directly connected. However, for branch boxes with a direct connection, it's still impossible to determine their hierarchical relationship.

[0120] In some embodiments of the present invention, the low voltage area topology identification method further includes:

[0121] S321: Determine whether there is a branch box with a direct connection relationship in the first connection relationship.

[0122] S322: If it exists, a high-frequency signal is sent to the low-voltage substation via a power carrier, and the signal strength of the branch box with a direct connection is obtained.

[0123] S323: Determine the superior-subordinate relationship of the branch boxes with a direct connection relationship according to the signal strength, and update the first connection relationship according to the superior-subordinate relationship.

[0124] In this embodiment, the relationship between the signal strength attenuation and the distance between the branch boxes can accurately determine the superior-subordinate relationship of the branch boxes with a direct connection relationship and update it to the first connection relationship, thereby ensuring the accuracy of the substation topology structure.

[0125] In some embodiments of the present invention, it is first determined whether there is a missed branch box. If not, it is determined whether there is a branch box with a direct connection relationship. If not, the second connection relationship of the meter box is directly determined.

[0126] In some embodiments of the present invention, as shown in FIG4 , determining the second connection relationship of the meter boxes based on the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box includes:

[0127] S41, for each last-level branch box, construct a linear relationship between the last-level branch box and all meter boxes based on the low-dimensional power data of the last-level branch box and the low-dimensional power data of the meter box, convert the linear relationship into a convex optimization model, and solve the convex optimization model to obtain the meter box connected to the last-level branch box.

[0128] S42: Obtain a second connection relationship between the meter boxes according to the meter boxes connected to each last-stage branch box.

[0129] In this embodiment, by converting the linear relationship into a convex optimization model, the amount of calculation can be reduced and the calculation efficiency can be improved.

[0130] In some embodiments of the present invention, converting the linear relationship into a convex optimization model includes:

[0131] S411, converting the linear relationship into an optimization model with the minimum solution error.

[0132] S412, replacing the L0 norm in the optimization model with the L2 norm to obtain a convex optimization model.

[0133] In this embodiment, by converting the linear relationship into an optimization model with the minimum solution error, an accurate topological relationship can be obtained when the noise interference β is unknown and there is a measurement error.

[0134] In some embodiments of the present invention, the linear relationship is:

[0135] The optimization model is:

[0136] st x m,i ∈{0,1}

[0137] The convex optimization model is:

[0138] st x m,i ∈[0,1]

[0139] in, represents the low-dimensional power data of the i-th last-level branch box, T' represents the number of moments of low-dimensional power data, Represents the low-dimensional power data of M meter boxes, x=[x 1,i ,x 2,i ,…,x m,i ,…,x M,i ] T It represents the meter box that builds a linear relationship with the i-th last-level branch box. When the m-th meter box and the i-th last-level branch box have a connection relationship, x m,i =1, otherwise x m,i =0, β represents the error caused by random noise interference in the power measurement value. represents the L0 norm, represents the L2 norm.

[0140] Specifically, the convex optimization model can be used to obtain the optimal solution for x. However, the calculated x vector contains decimals. To clarify the connection relationship, the decimal values ​​in the x vector need to be converted into 0-1 values, that is:

[0141] At this time, the element value in the x vector can indicate the meter box connected to the i-th branch box.

[0142] In this embodiment, by introducing a 0-1 variable to represent the connection relationship between the meter box and the branch box, the calculation efficiency can be improved.

[0143] In some embodiments of the present invention, as shown in FIG5 , dimensionality reduction processing is performed on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box, including:

[0144] S21, calculating the Gaussian probability distribution matrix of the original power data.

[0145] Specifically, calculating the Gaussian probability distribution matrix of the original electric power data includes: calculating the similarity conditional probability between each pair of data in the original electric power data, and calculating the first joint probability based on the similarity conditional probability, and obtaining the Gaussian probability distribution matrix of the original electric power data based on the first joint probability.

[0146] S22, constructing low-dimensional data, initializing the low-dimensional data, and calculating a t-distribution matrix of the low-dimensional data, wherein the number of moments of the low-dimensional data is smaller than the number of moments of the original power data.

[0147] Specifically, calculating the t-distribution matrix of the low-dimensional data includes: calculating the second joint probability of the low-dimensional data through the t-distribution function, wherein the initial value of the low-dimensional data is obtained by random initialization; and obtaining the t-distribution matrix of the low-dimensional data according to the second joint probability.

[0148] S23, calculates the KL divergence (Kullback-Leibler divergence, relative entropy) between the Gaussian probability distribution matrix and the t distribution matrix as a loss function.

[0149] S24, solving the low-dimensional data using the gradient descent method based on the loss function to obtain the low-dimensional power data of the branch box and the low-dimensional power data of the meter box.

[0150] In this embodiment, the t-SNE (t-distributed Stochastic Neighbor Embedding) algorithm is used to reduce the dimensionality of the original power data, thereby amplifying the differences between the data and weakening the impact of noise, providing high-quality low-dimensional power data for the subsequent topology recognition algorithm.

[0151] In some embodiments of the present invention, solving low-dimensional data using a gradient descent method based on a loss function includes:

[0152] S241, according to the loss function, calculate the low-dimensional data by the gradient descent method to obtain the gradient function.

[0153] S242: Obtain a gradient vector according to the gradient function, and update the low-dimensional data according to the gradient vector.

[0154] S243, determining whether the updated low-dimensional data meets the iteration end condition.

[0155] S244: If not reached, return to the step of calculating the t distribution matrix of the low-dimensional data.

[0156] S245: If reached, the current low-dimensional data is used as low-dimensional power data.

[0157] Specifically, the loss function is calculated by the following formula:

[0158] Among them, C i Represents the loss function of the i-th device to be identified.

[0159] It should be noted that the loss function C i The smaller , the more consistent the relative distribution of the i-th device to be identified with other devices before and after dimensionality reduction. That is, low-dimensional power data can better replace the original power data for topology identification.

[0160] The gradient function is calculated using the following formula:

[0161] The gradient vector is expressed as:

[0162] It should be noted that the gradient is calculated iteratively step by step, and the low-dimensional data is updated according to the obtained gradient. When the iteration condition is met, the low-dimensional power data can be obtained.

[0163] Iterative calculation is performed using the following formula:

[0164] Among them, q t represents the result of the t-th iteration of low-dimensional electrical data, η represents the learning rate, and α represents the momentum factor.

[0165] It should be noted that, when the iterative formula is solved repeatedly, When it is less than or equal to the set threshold, it is judged that the updated low-dimensional data meets the iteration end condition.

[0166] In some embodiments of the present invention, the similarity conditional probability is calculated using the following formula:

[0167] in, Indicates the raw power data. represents the time series data of the power of the i-th device to be identified, D represents the total number of devices to be identified, P j|i represents the similarity conditional probability of the i-th device to be identified being distributed around the j-th device to be identified, σ i Indicates The variance of a normal distribution centered at .

[0168] The first joint probability is calculated by the following formula:

[0169] Among them, p ij It represents the probability similarity between the i-th device to be identified and the j-th device to be identified.

[0170] The Gaussian probability distribution matrix is ​​expressed as: P = [p1, p2, ... p i ,…,p D ], where p i =[p i1 ,p i2 ,…,p iw ,…,p iD ], w≠i.

[0171] The second joint probability is calculated by the following formula:

[0172] Where Q = [q1;q2;…q i ;…;q D ]∈R D×T′ represents low-dimensional data, q i ∈R 1×T′ represents the low-dimensional data of the i-th device to be identified, and T' represents the number of moments of the low-dimensional data.

[0173] It should be noted that the low-dimensional data Q 0 It can be obtained by randomly initializing the matrix and then iteratively updating it in subsequent steps.

[0174] The t distribution matrix is ​​expressed as: R = [r1, r2, ..., r i ,…,r D ], where r i =[r i1 ,r i2 ,…r iv ,…,r iD ], v≠i.

[0175] Corresponding to the above embodiment, the present invention further proposes a low-voltage substation topology identification device.

[0176] FIG6 is a schematic structural diagram of a low-voltage substation topology identification device according to an embodiment of the present invention.

[0177] As shown in FIG6 , the low voltage substation topology identification device according to an embodiment of the present invention includes: a first acquisition module 10 , a data processing module 20 , a first determination module 30 , a second determination module 40 and a third determination module 50 . The first acquisition module 10 is used to obtain the power time series data of each device to be identified in the low-voltage substation to obtain the original power data; the data processing module 20 is used to perform dimensionality reduction processing on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box; the first determination module 30 is used to determine the substation total table based on the low-dimensional power data of the branch box, update the branch boxes whose connection relationships are to be determined, and group the branch boxes whose connection relationships are to be determined to obtain multiple combinations, and determine the target combination based on the low-dimensional power data of the branch boxes in each combination and the low-dimensional power data of the target branch box, as the next layer of the target branch box, re-update the branch boxes whose connection relationships are to be determined, until the last layer of the low-voltage substation branch box is reached, obtain the first connection relationship of the branch box, and obtain the last-level branch box, wherein the target branch box is the upper-level branch box of the branch box whose connection relationship is to be determined; the second determination module 40 is used to determine the second connection relationship of the meter box based on the low-dimensional power data of the last-level branch box and the low-dimensional power data of the meter box; the third determination module 50 is used to obtain the topological structure of the low-voltage substation based on the first connection relationship and the second connection relationship.

[0178] It should be noted that dimensionality reduction processing can reduce the dimension of data, reduce data redundancy, and weaken the impact of noise.

[0179] Specifically, the first connection relationship of the branch boxes can be determined based on the low-dimensional power data of the branch boxes and the principle of conservation of electrical energy. The principle of conservation of electrical energy states that within any time interval, the power of the measuring switch of the outgoing cabinet of a branch box and the measuring switch of its incoming cabinet are approximately equal at the same time.

[0180] More specifically, when determining the branch boxes in the first layer of the branch layer, all branch boxes except the substation master meter are considered as branch boxes whose connection relationships are to be determined. Subsequently, when determining the branch boxes in other layers of the branch layer, all branch boxes except the branch boxes whose connection relationships have been determined are considered as branch boxes whose connection relationships are to be determined.

[0181] As an example, first determine the substation total meter, take the substation total meter as the target branch box, and determine the meter box on the first layer in the branch layer; then, take the meter box on the first layer in the branch layer as the target branch box in turn, determine the meter box on the first layer in the branch layer and its corresponding relationship, and cycle in sequence until reaching the last layer of the low-voltage substation branch layer.

[0182] The low-voltage substation topology identification device of an embodiment of the present invention can reduce the interference caused by redundant information in high-dimensional data by performing dimensionality reduction processing on the original power data, and obtain the first connection relationship and the second connection relationship respectively through the low-dimensional power data, so as to accurately obtain the topological structure of the low-voltage substation.

[0183] In some embodiments of the present invention, when determining the substation total table based on the low-dimensional electricity data of the branch box, the first determination module 30 is used to: calculate the average value of the electricity data of each branch box based on the low-dimensional electricity data of the branch box, and use the branch box corresponding to the maximum average value as the substation total table.

[0184] In this embodiment, the first determining module 30 can quickly determine the substation master table in the branch box by calculating the average value of the power data of each branch box, sorting the data, and taking the branch box corresponding to the maximum average value as the substation master table.

[0185] In some embodiments of the present invention, when the first determining module 30 groups the branch boxes whose connection relationships are to be determined to obtain multiple combinations, it is configured to: randomly select 1, ..., e-1, e branch boxes from the branch boxes whose connection relationships are to be determined to group them, and obtain combinations, where K is the number of branch boxes whose connection relationships are to be determined.

[0186] In some embodiments of the present invention, the low-dimensional electricity data is time series data. When the first determination module 30 determines the target combination based on the low-dimensional electricity data of the branch boxes in each combination and the low-dimensional electricity data of the target branch box, it is used to: for each combination, add the electricity data at the same time in the low-dimensional electricity data of each branch box in the combination to obtain the cumulative electricity data of the combination; calculate the Pearson correlation coefficient of the cumulative electricity data of each combination and the low-dimensional electricity data of the target branch box respectively; and take the combination corresponding to the maximum Pearson correlation coefficient as the target combination.

[0187] Specifically, the Pearson correlation coefficient is calculated using the following formula:

[0188] in, represents the charge sequence of the mth branch box or target combination, cov(ε,τ) represents the covariance of ε and τ, and the symbol D(h) represents the variance of h.

[0189] As an example, assume that the number of devices to be identified is D, where N represents the number of branch boxes and M represents the number of meter boxes. The low-dimensional power data Q at the branch layer is B Expressed as:

[0190] in, is the low-dimensional power data of the i-th branch box.

[0191] The power sequence of each branch box Take the average value and record it as B1, B2,…, B N , and sort them from large to small according to the mean value, take the branch box corresponding to the maximum mean value as the substation area summary table of the substation layer, and then judge the connection relationship of the branch boxes at each level in the branch layer in turn.

[0192] In this embodiment, the first determination module 30 uses the Pearson correlation coefficient ρ (-1≤ρ≤1) to measure the similarity between the target combination and the target branch box. The larger the ρ, the more likely the target combination and the target branch box are parent-child nodes. Therefore, the above method can sequentially determine the connection relationship between each branch box in the branch layer.

[0193] In some embodiments of the present invention, the low-voltage substation topology identification device also includes: a first judgment module, used to determine whether there is a leaky branch box in the first connection relationship; a first calculation module, used to calculate the Pearson correlation coefficient of the low-dimensional power data of the leaky branch box and the cumulative power sequence corresponding to each target combination when there is a leaky branch box; a fourth determination module, used to determine the position of the leaky branch box in the first connection relationship according to the target combination corresponding to the maximum Pearson correlation coefficient, and update the first connection relationship according to the position.

[0194] The missing connection branch box may be connected between the target combination and the target branch box previously corresponding to the target combination.

[0195] In this embodiment, by acquiring all the leaked branch boxes, re-determining the connection relationship of the leaked branch boxes, and updating the connection relationship to the first connection relationship, the integrity of the substation topology structure can be ensured.

[0196] In some embodiments of the present invention, the low-voltage substation topology identification device also includes: a second judgment module, used to judge whether there is a branch box with a direct connection relationship in the first connection relationship; a second acquisition module, used to send a high-frequency signal to the low-voltage substation through the power carrier when there is a branch box with a direct connection relationship, and obtain the signal strength of the branch box with a direct connection relationship; a fifth determination module, used to determine the superior-subordinate relationship of the branch box with a direct connection relationship based on the signal strength, and update the first connection relationship based on the superior-subordinate relationship.

[0197] In this embodiment, the relationship between the signal strength attenuation and the distance between the branch boxes can accurately determine the superior-subordinate relationship of the branch boxes with a direct connection relationship and update it to the first connection relationship, thereby ensuring the accuracy of the substation topology structure.

[0198] In some embodiments of the present invention, when the second determination module 40 determines the second connection relationship of the meter box based on the low-dimensional power data of the last-level branch box and the low-dimensional power data of the meter box, it is used to: for each last-level branch box, construct a linear relationship between the last-level branch box and all the meter boxes based on the low-dimensional power data of the last-level branch box and the low-dimensional power data of the meter box, and convert the linear relationship into a convex optimization model, and solve the convex optimization model to obtain the meter box connected to the last-level branch box; according to the meter boxes connected to each last-level branch box, obtain the second connection relationship of the meter box.

[0199] In this embodiment, the second determination module 40 can reduce the amount of calculation and improve the calculation efficiency by converting the linear relationship into a convex optimization model.

[0200] In some embodiments of the present invention, when converting the linear relationship into a convex optimization model, the second determination module 40 is used to: convert the linear relationship into an optimization model with the smallest solution error; and replace the L0 norm in the optimization model with the L2 norm to obtain a convex optimization model.

[0201] In this embodiment, the second determination module 40 converts the linear relationship into an optimization model with the minimum solution error, so as to obtain an accurate topological relationship when the noise interference β is unknown and there is a measurement error.

[0202] In some embodiments of the present invention, the linear relationship is:

[0203] The optimization model is:

[0204] st x m,i ∈{0,1}

[0205] The convex optimization model is:

[0206] st x m,i ∈[0,1]

[0207] in, represents the low-dimensional power data of the i-th last-level branch box, T' represents the number of moments of low-dimensional power data, Represents the low-dimensional power data of M meter boxes, x=[x 1,i ,x 2,i ,…,x m,i ,…,x M,i ] T It represents the meter box that builds a linear relationship with the i-th last-level branch box. When the m-th meter box and the i-th last-level branch box have a connection relationship, x m,i =1, otherwise x m,i=0, β represents the error caused by random noise interference in the power measurement value. represents the L0 norm, represents the L2 norm.

[0208] Specifically, the second determination module 40 can obtain the optimal solution of x through the convex optimization model. However, the calculated x vector elements contain decimals. In order to clarify the connection relationship, the decimal values ​​in the x vector need to be converted into 0-1 values, that is:

[0209] At this time, the element value in the x vector can indicate the meter box connected to the i-th branch box.

[0210] In this embodiment, the second determining module 40 can improve calculation efficiency by introducing a 0-1 variable to represent the connection relationship between the meter box and the branch box.

[0211] In some embodiments of the present invention, when the data processing module 20 performs dimensionality reduction processing on the original electric power data to obtain low-dimensional electric power data of the branch box and low-dimensional electric power data of the meter box, it is used to: calculate the Gaussian probability distribution matrix of the original electric power data; construct low-dimensional data, initialize the low-dimensional data, and calculate the t-distribution matrix of the low-dimensional data, wherein the number of moments of the low-dimensional data is less than the number of moments of the original electric power data; calculate the KL divergence of the Gaussian probability distribution matrix and the t-distribution matrix as a loss function; solve the low-dimensional data using the gradient descent method based on the loss function to obtain the low-dimensional electric power data of the branch box and the low-dimensional electric power data of the meter box.

[0212] In this embodiment, the data processing module 20 reduces the dimensionality of the original power data by using the t-SNE algorithm, thereby amplifying the differences between the data and weakening the influence of noise, providing high-quality low-dimensional power data for the subsequent topology recognition algorithm.

[0213] In some embodiments of the present invention, when calculating the Gaussian probability distribution matrix of the original electric power data, the data processing module 20 is used to: calculate the similarity conditional probability between each pair of data in the original electric power data, and calculate the first joint probability based on the similarity conditional probability, and obtain the Gaussian probability distribution matrix of the original electric power data based on the first joint probability; when calculating the t distribution matrix of the low-dimensional data, the data processing module 20 is used to: calculate the second joint probability of the low-dimensional data through the t distribution function, wherein the initial value of the low-dimensional data is obtained by random initialization; and obtain the t distribution matrix of the low-dimensional data based on the second joint probability.

[0214] In some embodiments of the present invention, when the data processing module 20 uses the gradient descent method to solve the low-dimensional data based on the loss function, it is used to: calculate the low-dimensional data by the gradient descent method according to the loss function to obtain the gradient function; obtain the gradient vector according to the gradient function, and update the low-dimensional data according to the gradient vector; determine whether the updated low-dimensional data meets the iteration end condition; if not, return to the step of calculating the t distribution matrix of the low-dimensional data; if so, use the current low-dimensional data as the low-dimensional electrical quantity data.

[0215] Specifically, the data processing module 20 calculates the similarity conditional probability using the following formula:

[0216] in, Indicates the raw power data. represents the time series data of the power of the i-th device to be identified, D represents the total number of devices to be identified, P j|i represents the similarity conditional probability of the i-th device to be identified being distributed around the j-th device to be identified, σ i Indicates The variance of a normal distribution centered at .

[0217] The data processing module 20 calculates the first joint probability using the following formula:

[0218] Among them, p ij It represents the probability similarity between the i-th device to be identified and the j-th device to be identified.

[0219] Gaussian probability distribution matrix is ​​expressed as: P = [p1, p2, ... p i ,…,p D ], where p i =[p i1 ,p i2 ,…,p iw ,…,p iD ], w≠i.

[0220] The data processing module 20 calculates the second joint probability using the following formula:

[0221] Where Q = [q1;q2;…q i ;…;q D ]∈R D×T′ represents low-dimensional data, q i ∈R 1×T′ represents the low-dimensional data of the i-th device to be identified, and T' represents the number of moments of the low-dimensional data.

[0222] The t distribution matrix is ​​expressed as: R = [r1, r2, ..., ri ,…,r D ], where r i =[r i1 ,r i2 ,…r iv ,…,r iD ], v≠i.

[0223] More specifically, the data processing module 20 calculates the loss function using the following formula:

[0224] Among them, C i Represents the loss function of the i-th device to be identified.

[0225] The data processing module 20 calculates the gradient function using the following formula:

[0226] The gradient vector is expressed as:

[0227] More specifically, the data processing module 20 performs iterative calculations using the following formula:

[0228] Among them, q t represents the result of the t-th iteration of low-dimensional electrical data, η represents the learning rate, and α represents the momentum factor.

[0229] Corresponding to the low-voltage substation topology identification method of the above embodiment, the present invention further proposes a low-voltage substation topology identification system.

[0230] As shown in Figure 7, the low-voltage area topology identification system includes: a master station, a fusion terminal, a CCO (Central Coordinator), an LTU (Line Terminal Unit) and a STA (Station), where one STA is equivalent to one electricity meter.

[0231] The operation process of the low-voltage area topology identification system includes:

[0232] A1: The master station initiates a network-wide time calibration command, calibrating the converged terminal, CCO, LTU, and STA in sequence.

[0233] Specifically, the master station starts the synchronous clock, integrates the terminal synchronous system clock, the CCO synchronous system clock, the LTU synchronous network clock, and the STA synchronous network clock.

[0234] A2: The master station initiates a topology identification command to collect power data of the LTU and STA respectively.

[0235] Specifically, after receiving the topology identification command, the fusion terminal sends an instruction to collect power data to the CCO; the CCO sends the instruction to collect power data to the LTU and STA respectively; the LTU and STA report their respective power data to the CCO respectively; the CCO sends the received power data to the fusion terminal; the fusion terminal reports the LTU / STA power data to the master station.

[0236] A3: The master station performs topology identification using the above-mentioned low-voltage area topology identification method.

[0237] A4: The master station outputs the topology identification result, and the topology identification process ends.

[0238] It should be noted that the low-voltage distribution network includes transformers, distribution room main meters, branch boxes, meter boxes and user meters. Among them, the intelligent fusion terminal in the distribution box serves as the core equipment of the distribution network, and communicates with the intelligent circuit breaker / branch-side branch monitoring terminal / intelligent measuring switch through HPLC (Highspeed Power Line Carrier, high-speed power line carrier communication) or HRF (Highspeed Radio Frequency, high-speed wireless communication) and other methods. The low-voltage distribution network has the functions of integrating distribution terminal equipment (collecting distribution data) and concentrator (collecting marketing data), and can be responsible for the collection and monitoring of information such as the main meter switch on the distribution transformer side, the LTU on the branch side, and the intelligent measuring switch on the user side. At the same time, the intelligent fusion terminal analyzes, calculates, and organizes the substation data and transmits it to the distribution automation master station. Based on the data collection and edge computing capabilities of the substation intelligent terminal, the full-domain information perception and physical topology identification of the low-voltage distribution network can be realized.

[0239] Corresponding to the above embodiment, the present invention further proposes a non-transitory computer-readable storage medium.

[0240] The non-transitory computer-readable storage medium of the embodiment of the present invention stores a computer program, which, when executed by a processor, implements the low-voltage substation topology identification method proposed in the above embodiment of the present invention.

[0241] According to the non-transitory computer-readable storage medium of an embodiment of the present invention, by executing the computer program stored therein, the topological structure of the low-voltage substation can be accurately obtained, thereby generally improving the overall recognition accuracy.

[0242] Corresponding to the above embodiment, the present invention further provides a chip device.

[0243] FIG8 is a schematic structural diagram of a chip device according to an embodiment of the present invention.

[0244] As shown in Figure 8, chip device 500 includes a processor 501 and a memory 503. Processor 501 and memory 503 are connected, for example, via a bus 502. Optionally, chip device 500 may also include a transceiver 504. It should be noted that in practical applications, the number of transceivers 504 is not limited to one, and the structure of chip device 500 does not constitute a limitation on the embodiments of the present invention.

[0245] The processor 501 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The processor 501 may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the present disclosure. The processor 501 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0246] Bus 502 may include a path for transmitting information between the aforementioned components. Bus 502 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, for example. Bus 502 may be divided into an address bus, a data bus, a control bus, and so on. For ease of illustration, FIG8 shows only one thick line, but this does not indicate that there is only one bus or only one type of bus.

[0247] The memory 503 is used to store a computer program corresponding to the low-voltage substation topology identification method of the above embodiment of the present invention, and the computer program is controlled and executed by the processor 501. The processor 501 is used to execute the computer program stored in the memory 503 to implement the content shown in the above method embodiment.

[0248] The chip device 500 includes, but is not limited to, mobile terminals such as laptop computers and PADs (tablet computers), and fixed terminals such as desktop computers. The chip device 500 shown in FIG8 is merely an example and does not limit the functionality and scope of use of the embodiments of the present invention.

[0249] The chip device 500 according to the embodiment of the present invention can accurately obtain the topological structure of the low-voltage platform area.

[0250] It should be noted that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device), or in conjunction with such instruction execution system, apparatus, or device. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by an instruction execution system, apparatus, or device, or in conjunction with such instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wires (electronic device), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be 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 and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0251] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0252] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0253] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0254] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for identifying low voltage area topology, characterized in that: The method comprises: Obtaining the time series data of the power of each device to be identified in the low-voltage area to obtain the original power data; Performing dimensionality reduction processing on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box; Determine the substation general table according to the low-dimensional electricity data of the branch box, update the branch box with the connection relationship to be determined, and group the branch boxes with the connection relationship to be determined to obtain multiple combinations, and determine the target combination according to the low-dimensional electricity data of the branch boxes in each combination and the low-dimensional electricity data of the target branch box as the next layer of the target branch box, return to the step of updating the branch box with the connection relationship to be determined, until the last layer of the low-voltage substation branch box is divided, the first connection relationship of the branch box is obtained, and the last-level branch box is obtained, wherein the target branch box is the branch box of the upper layer of the branch box with the connection relationship to be determined; Determining a second connection relationship of the meter box according to the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box; According to the first connection relationship and the second connection relationship, a topological structure of the low-voltage area is obtained.

2. The low voltage area topology identification method according to claim 1, characterized in that: Determining the area master table according to the low-dimensional electricity data of the branch box includes: According to the low-dimensional power data of the branch boxes, the average value of the power data of each branch box is calculated, and the branch box corresponding to the maximum average value is used as the substation master table.

3. The low voltage area topology identification method according to claim 2, characterized in that: The branch boxes whose connection relationships are to be determined are grouped to obtain a plurality of combinations, including: From the branch boxes whose connection relationship is to be determined, 1, ..., e-1, e branch boxes are randomly selected for grouping, and the following is obtained: combinations, wherein K is the number of branch boxes whose connection relationship is to be determined.

4. The low voltage area topology identification method according to claim 2, characterized in that: The low-dimensional power data is time series data, and the target combination is determined according to the low-dimensional power data of the branch boxes in each combination and the low-dimensional power data of the target branch box, including: For each of the combinations, the power data at the same time in the low-dimensional power data of each branch box of the combination are added together to obtain the cumulative power data of the combination; Calculating the Pearson correlation coefficient of the accumulated power data of each combination and the low-dimensional power data of the target branch box respectively; The combination corresponding to the maximum Pearson correlation coefficient is taken as the target combination.

5. The low voltage area topology identification method according to claim 2, characterized in that: After obtaining the first connection relationship of the branch boxes, the method further includes: Determining whether there is a missing branch box in the first connection relationship; If so, respectively calculate the Pearson correlation coefficient between the low-dimensional power data of the leaky branch box and the cumulative power sequence corresponding to each target combination; According to the target combination corresponding to the maximum Pearson correlation coefficient, the position of the missing connection branch box in the first connection relationship is determined, and the first connection relationship is updated according to the position.

6. The low voltage area topology identification method according to claim 5, characterized in that: The method further comprises: Determine whether there is a branch box with a direct connection relationship in the first connection relationship; If it exists, a high-frequency signal is sent to the low-voltage area through a power carrier, and the signal strength of the branch box with a direct connection relationship is obtained; The superior-subordinate relationship of the branch boxes having a direct connection relationship is determined according to the signal strength, and the first connection relationship is updated according to the superior-subordinate relationship.

7. The low voltage area topology identification method according to claim 1, characterized in that: The determining the second connection relationship of the meter box according to the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box includes: For each of the final branch boxes, a linear relationship between the final branch box and all the meter boxes is constructed according to the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, and the linear relationship is converted into a convex optimization model, and the convex optimization model is solved to obtain the meter box connected to the final branch box; According to the meter boxes connected to each of the final-stage branch boxes, a second connection relationship of the meter boxes is obtained.

8. The low voltage area topology identification method according to claim 7, characterized in that: The converting the linear relationship into a convex optimization model comprises: Converting the linear relationship into an optimization model with the smallest solution error; The convex optimization model is obtained by replacing the L0 norm in the optimization model with the L2 norm.

9. The low voltage area topology identification method according to claim 8, characterized in that: The linear relationship is: The optimization model is: s.t.x m,i ∈{0,1} The convex optimization model is: s.t.x m,i ∈[0,1] in, represents the low-dimensional power data of the i-th terminal branch box, T' represents the time number of the low-dimensional power data, represents the low-dimensional power data of the M meter boxes, x=[x 1,i ,x 2,i ,…,x m,i ,…,x M,i ] T represents a meter box that establishes a linear relationship with the i-th terminal branch box. When the m-th meter box and the i-th terminal branch box have a connection relationship, x m,i =1, otherwise x m,i =0, β represents the error caused by random noise interference in the power measurement value, represents the L0 norm, represents the L2 norm.

10. The low voltage area topology identification method according to claim 1, characterized in that: The dimensionality reduction processing is performed on the original electric quantity data to obtain low-dimensional electric quantity data of the branch box and low-dimensional electric quantity data of the meter box, including: Calculate the Gaussian probability distribution matrix of the original electric quantity data; Constructing low-dimensional data, initializing the low-dimensional data, and calculating a t-distribution matrix of the low-dimensional data, wherein the number of moments of the low-dimensional data is less than the number of moments of the original power data; Calculating the KL divergence (Kullback-Leibler divergence, relative entropy) between the Gaussian probability distribution matrix and the t distribution matrix as a loss function; The low-dimensional data is solved based on the loss function using a gradient descent method to obtain low-dimensional electricity data of the branch box and low-dimensional electricity data of the meter box.

11. The low voltage area topology identification method according to claim 10, characterized in that: The calculating of the Gaussian probability distribution matrix of the original electric quantity data comprises: Calculating similarity conditional probabilities between pairs of data in the original electric quantity data, calculating a first joint probability according to the similarity conditional probabilities, and obtaining a Gaussian probability distribution matrix of the original electric quantity data according to the first joint probability; The calculating the t distribution matrix of the low-dimensional data comprises: Calculating a second joint probability of low-dimensional data by using a t distribution function, wherein an initial value of the low-dimensional data is obtained by random initialization; According to the second joint probability, a t distribution matrix of the low-dimensional data is obtained.

12. The low voltage area topology identification method according to claim 11, characterized in that: The step of solving the low-dimensional data by using a gradient descent method based on the loss function includes: According to the loss function, the low-dimensional data is calculated by the gradient descent method to obtain a gradient function; According to the gradient function, a gradient vector is obtained, and the low-dimensional data is updated according to the gradient vector; Determine whether the updated low-dimensional data meets the iteration end condition; If not reached, return to the step of calculating the t distribution matrix of the low-dimensional data; If it is reached, the current low-dimensional data is used as the low-dimensional electrical quantity data.

13. The low voltage area topology identification method according to claim 12, characterized in that: The similarity conditional probability is calculated by the following formula: in, represents the raw power data, represents the power time series data of the i-th device to be identified, D represents the total number of devices to be identified, P j|i represents the similarity conditional probability that the i-th device to be identified is distributed around the j-th device to be identified, σ i Indicates The variance of the normal distribution centered on The first joint probability is calculated by the following formula: Among them, p ij represents the probability similarity between the i-th device to be identified and the j-th device to be identified; The Gaussian probability distribution matrix is ​​expressed as: P = [p1, p2, ... p i ,…,p D ], where p i =[p i1 ,p i2 ,…,p iw ,…,p iD ], w≠i; The second joint probability is calculated by the following formula: Where Q = [q1; q2; ...q i ;…;q D ]∈R D×T′ represents the low-dimensional data, q i ∈R 1×T′ represents the low-dimensional data of the i-th device to be identified, and T' represents the time number of the low-dimensional data; The t distribution matrix is ​​expressed as: R = [r1, r2, ..., r i ,…,r D ], where r i =[r i1 ,r i2 ,…r iv ,…,r iD ], v≠i.

14. The low voltage area topology identification method according to claim 13, characterized in that: The loss function is calculated by the following formula: Among them, C i represents the loss function of the i-th device to be identified; The gradient function is calculated by the following formula: The gradient vector is expressed as:

15. The low voltage area topology identification method according to claim 14, characterized in that: Iterative calculation is performed using the following formula: Among them, q t represents the result of the tth iteration of low-dimensional electrical data, η represents the learning rate, and α represents the momentum factor.

16. A low voltage area topology identification device, characterized in that: The device comprises: The first acquisition module is used to acquire the power time series data of each device to be identified in the low-voltage area to obtain the original power data; A data processing module, used for performing dimensionality reduction processing on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box; A first determination module is used to determine the substation general table according to the low-dimensional electricity data of the branch box, update the branch box with the connection relationship to be determined, and group the branch boxes with the connection relationship to be determined to obtain multiple combinations, and determine the target combination according to the low-dimensional electricity data of the branch boxes in each combination and the low-dimensional electricity data of the target branch box as the next layer of the target branch box, re-update the branch boxes with the connection relationship to be determined until the last layer of the branch box in the low-voltage substation is reached, obtain the first connection relationship of the branch box, and obtain the last-level branch box, wherein the target branch box is the branch box of the upper layer of the branch box with the connection relationship to be determined; A second determining module, configured to determine a second connection relationship of the meter box according to the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box; The third determining module is used to obtain the topological structure of the low-voltage area according to the first connection relationship and the second connection relationship.

17. The low voltage area topology identification device according to claim 16, characterized in that: When determining the total table of the substation area according to the low-dimensional electricity data of the branch box, the first determination module is used to: According to the low-dimensional power data of the branch boxes, the average value of the power data of each branch box is calculated, and the branch box corresponding to the maximum average value is used as the substation master table.

18. The low voltage area topology identification device according to claim 17, characterized in that: When the first determining module groups the branch boxes whose connection relationships are to be determined to obtain a plurality of combinations, it is configured to: From the branch boxes whose connection relationship is to be determined, 1, ..., e-1, e branch boxes are randomly selected for grouping, and the following is obtained: combinations, wherein K is the number of branch boxes whose connection relationship is to be determined.

19. The low voltage area topology identification device according to claim 17, characterized in that: The low-dimensional power data is time series data. When the first determination module determines the target combination according to the low-dimensional power data of the branch boxes in each combination and the low-dimensional power data of the target branch box, it is used to: For each of the combinations, the power data at the same time in the low-dimensional power data of each branch box of the combination are added together to obtain the cumulative power data of the combination; Calculating the Pearson correlation coefficient of the accumulated power data of each combination and the low-dimensional power data of the target branch box respectively; The combination corresponding to the maximum Pearson correlation coefficient is taken as the target combination.

20. The low voltage area topology identification device according to claim 17, characterized in that: The device also includes: A first judging module, used for judging whether there is a missing branch box in the first connection relationship; A first calculation module is used for respectively calculating the Pearson correlation coefficient between the low-dimensional power data of the missing branch box and the accumulated power sequence corresponding to each target combination when there is a missing branch box; The fourth determination module is used to determine the position of the missing connection branch box in the first connection relationship according to the target combination corresponding to the maximum Pearson correlation coefficient, and update the first connection relationship according to the position.

21. The low voltage area topology identification device according to claim 20, characterized in that: The device also includes: A second judgment module is used to judge whether there is a branch box with a direct connection relationship in the first connection relationship; A second acquisition module is used to send a high-frequency signal to the low-voltage area through a power carrier when there is a branch box with a direct connection relationship, and obtain the signal strength of the branch box with a direct connection relationship; The fifth determining module is used to determine the superior-subordinate relationship of the branch boxes with the direct connection relationship according to the signal strength, and update the first connection relationship according to the superior-subordinate relationship.

22. The low voltage area topology identification device according to claim 16, characterized in that: When determining the second connection relationship of the meter box according to the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, the second determination module is used to: For each of the final branch boxes, a linear relationship between the final branch box and all the meter boxes is constructed according to the low-dimensional power data of the final branch box and the low-dimensional power data of the meter box, and the linear relationship is converted into a convex optimization model, and the convex optimization model is solved to obtain the meter box connected to the final branch box; According to the meter boxes connected to each of the final-stage branch boxes, a second connection relationship of the meter boxes is obtained.

23. The low voltage area topology identification device according to claim 22, characterized in that: When converting the linear relationship into a convex optimization model, the second determination module is used to: Converting the linear relationship into an optimization model with the smallest solution error; The convex optimization model is obtained by replacing the L0 norm in the optimization model with the L2 norm.

24. The low voltage area topology identification device according to claim 23, characterized in that: The linear relationship is: The optimization model is: s.t.x m,i ∈{0,1} The convex optimization model is: s.t.x m,i ∈[0,1] in, represents the low-dimensional power data of the i-th terminal branch box, T' represents the time number of the low-dimensional power data, represents the low-dimensional power data of the M meter boxes, x=[x 1,i ,x 2,i ,…,x m,i ,…,x M,i ] T represents a meter box that establishes a linear relationship with the i-th terminal branch box. When the m-th meter box and the i-th terminal branch box have a connection relationship, x m,i =1, otherwise x m,i =0, β represents the error caused by random noise interference in the power measurement value, represents the L0 norm, represents the L2 norm.

25. The low voltage area topology identification device according to claim 16, characterized in that: When the data processing module performs dimensionality reduction processing on the original power data to obtain low-dimensional power data of the branch box and low-dimensional power data of the meter box, it is used to: Calculate the Gaussian probability distribution matrix of the original electric quantity data; Constructing low-dimensional data, initializing the low-dimensional data, and calculating a t-distribution matrix of the low-dimensional data, wherein the number of moments of the low-dimensional data is less than the number of moments of the original power data; Calculate the KL divergence of the Gaussian probability distribution matrix and the t distribution matrix as a loss function; The low-dimensional data is solved based on the loss function using a gradient descent method to obtain low-dimensional electricity data of the branch box and low-dimensional electricity data of the meter box.

26. The low voltage area topology identification device according to claim 25, characterized in that: When calculating the Gaussian probability distribution matrix of the original electric quantity data, the data processing module is used to: Calculating similarity conditional probabilities between pairs of data in the original electric quantity data, calculating a first joint probability according to the similarity conditional probabilities, and obtaining a Gaussian probability distribution matrix of the original electric quantity data according to the first joint probability; When calculating the t distribution matrix of the low-dimensional data, the data processing module is used to: Calculating a second joint probability of low-dimensional data by using a t distribution function, wherein an initial value of the low-dimensional data is obtained by random initialization; According to the second joint probability, a t distribution matrix of the low-dimensional data is obtained.

27. The low voltage area topology identification device according to claim 26, characterized in that: When the data processing module solves the low-dimensional data using the gradient descent method based on the loss function, it is used to: According to the loss function, the low-dimensional data is calculated by the gradient descent method to obtain a gradient function; According to the gradient function, a gradient vector is obtained, and the low-dimensional data is updated according to the gradient vector; Determine whether the updated low-dimensional data meets the iteration end condition; If not reached, return to the step of calculating the t distribution matrix of the low-dimensional data; If it is reached, the current low-dimensional data is used as the low-dimensional electrical quantity data.

28. The low voltage area topology identification device according to claim 27, characterized in that: The data processing module calculates the similarity conditional probability by the following formula: in, represents the raw power data, represents the power time series data of the i-th device to be identified, D represents the total number of devices to be identified, P j|i represents the similarity conditional probability that the i-th device to be identified is distributed around the j-th device to be identified, σ i Indicates The variance of the normal distribution centered on The data processing module calculates the first joint probability by the following formula: Among them, p ij represents the probability similarity between the i-th device to be identified and the j-th device to be identified; The Gaussian probability distribution matrix is ​​expressed as: P = [p1, p2, ... p i ,…,p D ], where p i =[p i1 ,p i2 ,…,p iw ,…,p iD ], w≠i; The data processing module calculates the second joint probability by the following formula: Where Q = [q1; q2; ...q i ;…;q D ]∈R D×T′ represents the low-dimensional data, q i ∈R 1×T′ represents the low-dimensional data of the i-th device to be identified, and T' represents the time number of the low-dimensional data; The t distribution matrix is ​​expressed as: R = [r1, r2, ..., r i ,…,r D ], where r i =[r i1 ,r i2 ,…r iv ,…,r iD ], v≠i.

29. The low voltage area topology identification device according to claim 28, characterized in that: The data processing module calculates the loss function using the following formula: Among them, C i represents the loss function of the i-th device to be identified; The data processing module calculates the gradient function by the following formula: The gradient vector is expressed as:

30. The low voltage area topology identification device according to claim 29, characterized in that: The data processing module performs iterative calculations using the following formula: Among them, q t represents the result of the tth iteration of low-dimensional electrical data, η represents the learning rate, and α represents the momentum factor.

31. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the low-voltage area topology identification method as described in any one of claims 1 to 15 is implemented.

32. A chip device comprising a memory, a processor and a computer program stored in the memory, characterized in that: When the computer program is executed by the processor, the low-voltage substation topology identification method according to any one of claims 1 to 15 is implemented.

Images