Self-supervised Low-voltage Substation Topology Identification Method Based on Stacked Ridge Regression

Abstract

The present invention provides a self-supervised low-voltage substation topology identification method based on stacked ridge regression, which relates to the technical field of distribution network management. It includes step S1: the energy controller executes the networking command and broadcasts the time calibration of the substation measuring equipment; step S2: determines whether the substation topology has changed, and if so, executes step S3; if not, calculates whether the line loss rate of each level of the substation exceeds the limit, and if so, executes step S3, otherwise terminates the execution of the current identification method; step S3: the energy controller collects 96 points of forward active power data of the substation measuring equipment; step S4: establishes a stacked ridge regression model, and sorts out the topological relationship step by step; step S5: determines whether the substation topology structure needs to be updated and performs corresponding execution. The present invention can realize the topological identification of low-voltage substations, and solves the problem that the existing low-voltage substation topology identification methods all have limitations.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network management, and in particular to a self-supervised low-voltage substation topology recognition method based on stacked ridge regression. Background Art

[0002] In recent years, with the significant increase in requirements for stability and low losses, loss reduction has become one of the key tasks of power supply companies. Loss reduction is closely related to topology identification. By improving the quality and efficiency of topology identification, loss reduction goals can be achieved more effectively. Currently, the topology sorting methods for 10kV distribution networks are relatively clear, but at the low-voltage substation level, there are still many cases where the substation topology is unclear and the relationship between households and transformers is very chaotic. Based on this, for high-loss substations, the chaotic topology structure will greatly increase the workload of line loss investigation; problems such as branch node power outages and line faults cannot be quickly and effectively located, which in turn affects the efficiency of emergency repairs. Therefore, the unclear topology problem in low-voltage substations will seriously affect the advancement of lean line loss management and the efficiency of detection and emergency repairs when substation faults occur.

[0003] In the existing technology, a considerable number of power supply companies still use traditional manual topology combing methods, such as power outage differentiation and on-site manual line combing, which are inefficient and have limited effectiveness. With the development of technology, some topology identification methods using computer technology have also emerged. Among them, the mainstream solutions include the correlation method based on substation information and the method based on characteristic current signals. For example, patent publication number CN110727662A provides a low-voltage substation user phase identification method and system based on correlation analysis, and patent publication number CN114142464A provides a low-voltage substation topology identification method and device based on characteristic current signal detection. However, the correlation method based on substation information is more dependent on the efficiency and accuracy of information collection, and the identification cycle is long. It is less effective for identifying common zero, coupled, and back-to-back substations. The method based on characteristic current signals requires additional equipment, has high overall operating costs, and requires the injection of high-frequency signals into the power lines, which will have a certain impact on the stability of the power grid.

[0004] Therefore, it is necessary to study a safe, reliable, efficient and economical low-voltage substation topology identification method. Summary of the Invention

[0005] The purpose of the present invention is to provide a self-supervised low-voltage area topology identification method based on stacked ridge regression to solve the problem mentioned in the above background technology that the existing low-voltage area topology identification methods have limitations.

[0006] The present invention is achieved by adopting the following technical solutions:

[0007] A self-supervised low-voltage station topology recognition method based on stacked ridge regression includes the following steps:

[0008] Step S1: The energy controller in the low-voltage area executes the networking command and broadcasts time calibration to the area's measuring equipment;

[0009] Step S2: Determine whether the area topology has changed based on the networking result. If so, directly execute step S3. If not, calculate whether the line loss rate at each level in the area exceeds the limit. If so, execute step S3. If not, terminate the execution of the current identification method.

[0010] Step S3: The energy controller collects 96 points of forward active power data from the measurement equipment in the substation;

[0011] Step S4: Establish a stacked ridge regression model, calculate the topological results based on the model, and sort out the topological relationships step by step;

[0012] Step S5: Calculate the topology results multiple times and compare them with the existing archive information of the substation area to determine whether the substation area topology structure needs to be updated and perform corresponding operations.

[0013] Furthermore, in step S1: broadcast time calibration is performed to make the clock synchronization error between the measurement devices in the station area less than 2s.

[0014] Furthermore, in the step S2: comparing whether there is any discrepancy between the networking result and the existing archive information of the substation area, if there is a discrepancy, it means that the substation area topology has changed.

[0015] Furthermore, the evaluation index of whether there is a discrepancy includes:

[0016] a. The network results contain additional meter records;

[0017] b. The network configuration results contain deleted meter records;

[0018] c. The networking results contain a record of meter replacement.

[0019] Furthermore, in step S2: if the absolute value of the line loss rate of each level in the substation exceeds 70%, it means that the limit is exceeded. The formula for calculating the line loss rate of each level in the substation is:

[0020] ,

[0021] Where, represents the line loss rate of the b-th level node, Indicates the b The parent node of each level node freezes the power data of the day. Indicates the b The corresponding node in the leveli Freeze the power data of each child node on that day.

[0022] Furthermore, in step S4: a stacked ridge regression model is established based on the principle that the active power of parent and child nodes at all levels satisfies the power conservation principle, the regression coefficient is calculated and solved, and the child nodes whose coefficient values exceed the preset value and whose average power is relatively significant are assigned to the corresponding parent nodes, so as to sort out the topological relationship step by step.

[0023] Furthermore, the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels in the substation measurement equipment is sorted out, which specifically includes the following sub-steps:

[0024] Step S4-A1: Initialize and set the level data parameters b =1, parent node number parameter j =1, the child node set is initialized to empty ;

[0025] Step S4-A2: Create the b Level j The stacked ridge regression model of parent-child nodes is initialized with the number of stacking times. t =1; initialize the number of child nodes is the maximum number that can be reached, that is, the number of all intelligent circuit breakers excluding the parent node;

[0026] Step S4-A3: Calculate the b Level j Parent node t Coefficient vector of the subsolution , the formula is,

[0027] ,

[0028] Where, is the active power matrix composed of the current intelligent circuit breaker sub-nodes, N Indicates the number of data points; Represents the current parent node's remaining active power data, that is, the original active power of the parent node minus the algebraic sum of the active powers of the determined child nodes; represents the identity matrix;

[0029] Step S4-A4: Select coefficient vector The set of child nodes corresponding to the coefficients exceeding 0.85 in the calculation is used to calculate the average active power of each child node; the set of child nodes whose average active power is greater than the mean of the average active power of all child nodes is retained. To determine the child node, ; Update the parent-child node data, that is, the remaining active power data at the current parent node Subtract the child node set from The active power algebraic sum of the active power matrix composed of the current intelligent circuit breaker sub-nodes Remove the child node set The data sequence represented;

[0030] Step S4-A5: If the coefficient vector If the number of stacking times exceeds 0.85 is not more than 1, then execute step S4-A4 and then execute step S4-A6; otherwise, execute the stacking times t Do incremental operation , repeat steps S4-A3 to S4-A4;

[0031] Step S4-A6: If the level data parameter b =1, then execute step S4-A7; otherwise, the number of parent nodes parameter j Do incremental operation , the current parent node remaining active power data Updated to j Repeat steps S4-A2 to S4-A5 to sort out the data of the parent node. j The topology of each parent node is sorted out until all parent nodes of the current level are sorted out;

[0032] Step S4-A7: Level data parameters b Do incremental operation , reset the number of parent nodes at the current level j =1, child node set , record the child nodes of the current level as the parent nodes of the next level, repeat steps S4-A2 to S4-A6 until all intelligent circuit breakers are divided, and output the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels .

[0033] Furthermore, the topological relationship between the terminal intelligent circuit breaker as the meter box layer in the substation measurement equipment and all user meters is sorted out, which specifically includes the following sub-steps:

[0034] Step S4-B1: Initialize the child node set to empty, i.e. ;

[0035] Step S4-B2: Build a stacked ridge regression model for all parent and child nodes and initialize the number of stacking times t =1;

[0036] Step S4-B3: Calculate the t The coefficient matrix of the subsolution , the calculation formula is,

[0037] ,

[0038] Where, is the active power matrix composed of all user tables, U Indicates the number of user tables. Indicates the t The residual active power matrix composed of all terminal intelligent circuit breakers in the second calculation is the original active power of each parent node minus the algebraic sum of the active power of its subordinate child nodes that have been determined;

[0039] Step S4-B4: Determine the pre-assignment relationship between the meter box and the user according to the maximum coefficient principle, that is, the pre-assignment number of the terminal intelligent circuit breaker to which the user meter belongs is ; Moreover, in each parent node, a set of child nodes is retained whose average active power is greater than the mean of the average active power of all corresponding child nodes To determine the child node, update the child node set , update the parent-child node data, that is, the residual active power matrix composed of all terminal intelligent circuit breakers Subtract the corresponding child node set from The active power algebraic sum of all user tables is the active power matrix Remove the child node set The data sequence represented;

[0040] Step S4-B5: Stacking times t Do incremental operation , repeat steps S4-B3 to S4-B4, if the active power matrix composed of all user tables is The number of remaining columns does not exceed the number of terminal intelligent circuit breakers J , then retain the result and output the topological relationship between the terminal intelligent circuit breaker and all user tables.

[0041] Furthermore, in step S5: if the topology result is consistent with the existing archival information of the substation, the user table corresponding to the regression coefficient exceeding the preset value in the hierarchical node where the line loss rate exceeds the warning value is recorded, and it is reported as an abnormal event; otherwise, the line loss rate of each level with the intelligent circuit breaker in the substation measuring equipment as the parent node is calculated, and the result with the lowest average line loss rate is reported.

[0042] Furthermore, the warning value is 70% and the preset value is 1.5.

[0043] The beneficial effects achieved by the present invention are:

[0044] By executing steps S1 to S5, the topology identification of the low-voltage substation can be achieved, wherein: by comparing the networking results of the energy actuator with the existing archival information of the substation, and according to the changes in the line loss rates at various levels of the substation, it is possible to self-supervise and judge whether to execute the topology identification command, which has strong adaptability; and, the calculation is based on the active power data, the calculation amount is small, there is no need to install signal injection equipment, and the measurement signal does not need to be uploaded to the main station, and can be analyzed on the local device, which is friendly to the safety of the power grid and has good engineering practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a schematic diagram of a flow chart of a topology identification method according to an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of the low-voltage area topology described as an example in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The technical solutions in the embodiments of the present invention will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present invention.

[0048] Example 1

[0049] Please refer to Figures 1 to 2 The present invention provides a self-supervised low-voltage station area topology identification method based on stacked ridge regression, comprising the following steps:

[0050] Step S1: The energy controller in the low-voltage area executes the networking command and broadcasts the time calibration to the area's measuring equipment. Specifically:

[0051] In this step, broadcast time calibration is performed to ensure that the clock synchronization error between the measurement equipment in the substation is less than 2s, thereby ensuring the clock synchronization of the measurement equipment in the substation; the measurement equipment in the substation includes smart circuit breakers and smart meters.

[0052] Step S2: Determine whether the area topology has changed based on the networking result. If so, directly execute step S3. If not, calculate whether the line loss rate of each level in the area exceeds the limit. If so, execute step S3. If not, terminate the execution of the current identification method.

[0053] In this step, the network configuration results are compared with the existing archived information of the substation to see if there are any discrepancies. If there are any discrepancies, it means that the substation topology has changed. The evaluation indicators for whether there are any discrepancies include:

[0054] a. The network results contain additional meter records;

[0055] b. The network configuration results contain deleted meter records;

[0056] c. The networking results contain a record of meter replacement.

[0057] If the absolute value of the line loss rate at each level in the substation exceeds 70%, it means that the limit is exceeded. The formula for calculating the line loss rate at each level in the substation is:

[0058] ,

[0059] Where, represents the line loss rate of the b-th level node, Indicates the b The parent node of each level node freezes the power data of the day. Indicates the b The corresponding node in the level i Freeze the power data of each child node on that day.

[0060] Step S3: The energy controller collects 96 points of forward active power data from the area's measuring equipment through HPLC on a daily basis.

[0061] Step S4: Establish a stacked ridge regression model, calculate the topological results based on the model, and sort out the topological relationships step by step. Specifically:

[0062] In this step, a stacked ridge regression model is established based on the principle that the active power of parent and child nodes at all levels satisfies the power conservation principle, the regression coefficient is calculated and solved, and the child nodes with coefficient values exceeding the preset value and relatively significant average power are assigned to the corresponding parent nodes to sort out the topological relationship step by step; this includes sorting out the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels in the substation measurement equipment, as well as sorting out the topological relationship between the terminal intelligent circuit breaker as the meter box layer in the substation measurement equipment and all user meters.

[0063] When sorting out the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels in the substation measurement equipment, the following sub-steps are specifically included:

[0064] Step S4-A1: Initialize and set the level data parameters b =1, parent node number parameter j =1, the child node set is initialized to empty ;

[0065] Step S4-A2: Create the b Level j The stacked ridge regression model of parent-child nodes is initialized with the number of stacking times. t =1; initialize the number of child nodes is the maximum number that can be reached, that is, the number of all intelligent circuit breakers excluding the parent node;

[0066] Step S4-A3: Calculate the b Levelj Parent node t Coefficient vector of the subsolution , the formula is,

[0067] ,

[0068] Where, is the active power matrix composed of the current intelligent circuit breaker sub-nodes, N Indicates the number of data points; Represents the current parent node's remaining active power data, that is, the original active power of the parent node minus the algebraic sum of the active powers of the determined child nodes; represents the identity matrix;

[0069] Step S4-A4: Select coefficient vector The set of child nodes corresponding to the coefficients exceeding 0.85 in the calculation is used to calculate the average active power of each child node; the set of child nodes whose average active power is greater than the mean of the average active power of all child nodes is retained. To determine the child node, ; Update the parent-child node data, that is, the remaining active power data at the current parent node Subtract the child node set from The active power algebraic sum of the active power matrix composed of the current intelligent circuit breaker sub-nodes Remove the child node set The data sequence represented;

[0070] Step S4-A5: If the coefficient vector If the number of stacking times exceeds 0.85 is not more than 1, then execute step S4-A4 and then execute step S4-A6; otherwise, execute the stacking times t Do incremental operation , repeat steps S4-A3 to S4-A4;

[0071] Step S4-A6: If the level data parameter b =1, then execute step S4-A7; otherwise, the number of parent nodes parameter j Do incremental operation , the current parent node remaining active power data Updated to j Repeat steps S4-A2 to S4-A5 to sort out the data of the parent node. j The topology of each parent node is sorted out until all parent nodes of the current level are sorted out;

[0072] Step S4-A7: Level data parameters b Do incremental operation , reset the number of parent nodes at the current levelj =1, child node set , record the child nodes of the current level as the parent nodes of the next level, repeat steps S4-A2 to S4-A6 until all intelligent circuit breakers are divided, and output the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels .

[0073] When sorting out the topological relationship between the terminal intelligent circuit breaker as the meter box layer in the substation measurement equipment and all user meters, the specific sub-steps include the following:

[0074] Step S4-B1: Initialize the child node set to empty, i.e. ;

[0075] Step S4-B2: Build a stacked ridge regression model for all parent and child nodes and initialize the number of stacking times t =1;

[0076] Step S4-B3: Calculate the t The coefficient matrix of the subsolution , the calculation formula is,

[0077] ,

[0078] Where, is the active power matrix composed of all user tables, U Indicates the number of user tables. Indicates the t The residual active power matrix composed of all terminal intelligent circuit breakers in the second calculation is the original active power of each parent node minus the algebraic sum of the active power of its subordinate child nodes that have been determined;

[0079] Step S4-B4: Determine the pre-assignment relationship between the meter box and the user according to the maximum coefficient principle, that is, the pre-assignment number of the terminal intelligent circuit breaker to which the user meter belongs is ; Moreover, in each parent node, a set of child nodes is retained whose average active power is greater than the mean of the average active power of all corresponding child nodes To determine the child node, update the child node set , update the parent-child node data, that is, the residual active power matrix composed of all terminal intelligent circuit breakers Subtract the corresponding child node set from The active power algebraic sum of all user tables is the active power matrix Remove the child node set The data sequence represented;

[0080] Step S4-B5: Stacking times t Do incremental operation , repeat steps S4-B3 to S4-B4, if the active power matrix composed of all user tables is The number of remaining columns does not exceed the number of terminal intelligent circuit breakers J , then retain the result and output the topological relationship between the terminal intelligent circuit breaker and all user tables.

[0081] Step S5: Calculate the topology results multiple times and compare them with the existing archive information of the substation to determine whether the substation topology needs to be updated and perform the corresponding operations.

[0082] In this step, if the topology result matches the existing information in the substation's archive, the user table corresponding to the regression coefficient exceeding the preset value in the hierarchical node where the line loss rate exceeds the warning value is recorded and reported as an abnormal event, indicating that the corresponding user table is abnormally using electricity or suspected of electricity theft. Otherwise, the line loss rate of each level is calculated, with the intelligent circuit breaker in the substation's measurement equipment as the parent node, and the result with the lowest average line loss rate is reported. The warning value is 70%, and the preset value is 1.5.

[0083] In summary, by executing steps S1 to S5, the topology identification of the low-voltage substation can be achieved. The topology identification method provided in this embodiment can self-supervise and judge whether to execute the topology identification command by comparing the networking results of the energy actuator with the existing archival information of the substation, and according to the changes in the line loss rate at each level of the substation, and has strong adaptability; and, based on the active power data, the calculation amount is small, and there is no need to install signal injection equipment. The measurement signal does not need to be uploaded to the main station, and can be analyzed on the local device. It is friendly to the safety of the power grid and has good engineering practicality.

[0084] It should be noted that the parts that are not described in detail or expanded in the above scheme are all existing technologies, which do not belong to the improvements made by the present invention on the existing technologies, nor do they fall within the scope of protection of the technical scheme of the present invention. Therefore, they will not be described in detail in this article.

[0085] Of course, the above contents are only preferred embodiments of the present invention and should not be considered as limiting the scope of the embodiments of the present invention. The present invention is not limited to the above examples. Equivalent changes and improvements made by ordinary technicians in this technical field within the essential scope of the present invention should all fall within the scope of the patent of the present invention.

Claims

1. A self-supervised low-voltage station topology identification method based on stacked ridge regression, characterized in that: The steps include: Step S1: The energy controller in the low-voltage area executes the networking command and broadcasts time calibration to the area's measuring equipment; Step S2: Determine whether the area topology has changed based on the networking result. If so, directly execute step S3. If not, calculate whether the line loss rate at each level in the area exceeds the limit. If so, execute step S3. If not, terminate the execution of the current identification method. Step S3: The energy controller collects 96 points of forward active power data from the measurement equipment in the substation; Step S4: Establish a stacked ridge regression model, calculate the topological results based on the model, and sort out the topological relationships step by step; Step S5: Calculate the topology results multiple times and compare them with the existing archive information of the substation area to determine whether the substation area topology structure needs to be updated and perform the corresponding operation; In step S4, a stacked ridge regression model is established based on the principle that the active power of parent and child nodes at all levels satisfies the power conservation principle, the regression coefficient is calculated and solved, and the child nodes with coefficient values exceeding the preset value and relatively significant average power are assigned to the corresponding parent nodes, so as to sort out the topological relationship step by step; this includes sorting out the topological relationship between the terminal intelligent circuit breaker as the meter box layer in the substation measurement equipment and all user meters, which specifically includes the following sub-steps: Step S4-B1: Initialize the child node set to empty, i.e. ; Step S4-B2: Build a stacked ridge regression model for all parent and child nodes and initialize the number of stacking times t =1; Step S4-B3: Calculate the t The coefficient matrix of the subsolution , the calculation formula is, ， Where, is the active power matrix composed of all user tables, U Indicates the number of user tables. Indicates the t The residual active power matrix composed of all terminal intelligent circuit breakers in the second calculation is the original active power of each parent node minus the algebraic sum of the active power of its subordinate child nodes that have been determined; Step S4-B4: Determine the pre-assignment relationship between the meter box and the user according to the maximum coefficient principle, that is, the pre-assignment number of the terminal intelligent circuit breaker to which the user meter belongs is ; Moreover, in each parent node, a set of child nodes is retained whose average active power is greater than the mean of the average active power of all corresponding child nodes To determine the child node, update the child node set , update the parent-child node data, that is, the residual active power matrix composed of all terminal intelligent circuit breakers Subtract the corresponding child node set from The active power algebraic sum of all user tables is the active power matrix Remove the child node set The data sequence represented; Step S4-B5: Stacking times t Do incremental operation , repeat steps S4-B3 to S4-B4, if the active power matrix composed of all user tables is The number of remaining columns does not exceed the number of terminal intelligent circuit breakers J , then retain the result and output the topological relationship between the terminal intelligent circuit breaker and all user tables.

2. The self-supervised low-voltage station area topology identification method based on stacked ridge regression according to claim 1 is characterized in that: In the step S1: broadcast time calibration is performed to make the clock synchronization error between the measuring devices in the station area less than 2s.

3. The self-supervised low-voltage station area topology identification method based on stacked ridge regression according to claim 1 is characterized in that: In the step S2: comparing whether there is any discrepancy between the networking result and the existing archive information of the substation area, if there is any discrepancy, it means that the substation area topology has changed.

4. The self-supervised low-voltage station area topology identification method based on stacked ridge regression according to claim 3 is characterized in that: The evaluation indicators for whether there is a discrepancy include: a. The network configuration results contain additional meter records; b. The network configuration results contain deleted meter records; c. The networking results contain a record of meter replacement.

5. The self-supervised low-voltage station area topology identification method based on stacked ridge regression according to claim 1 is characterized in that: In step S2: if the absolute value of the line loss rate of each level in the substation exceeds 70%, it means that the limit is exceeded. The formula for calculating the line loss rate of each level in the substation is: ， Where, represents the line loss rate of the b-th level node, Indicates the b The parent node of each level node freezes the power data of the day. Indicates the b The corresponding node in the level i Freeze the power data of each child node on that day.

6. The self-supervised low-voltage station topology identification method based on stacked ridge regression according to claim 1 is characterized in that: This involves sorting out the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels in the substation measurement equipment, specifically including the following sub-steps: Step S4-A1: Initialize and set the level data parameters b =1, parent node number parameter j =1, the child node set is initialized to empty ; Step S4-A2: Create the b Level j The stacked ridge regression model of parent-child nodes is initialized with the number of stacking times. t =1; initialize the number of child nodes is the maximum number that can be reached, that is, the number of all intelligent circuit breakers excluding the parent node; Step S4-A3: Calculate the b Level j Parent node t Coefficient vector of the subsolution , the formula is, ， Where, is the active power matrix composed of the current intelligent circuit breaker sub-nodes, N Indicates the number of data points; Represents the current remaining active power data of the parent node, that is, the original active power of the parent node minus the algebraic sum of the active powers of the determined child nodes; represents the identity matrix; Step S4-A4: Select coefficient vector The set of child nodes corresponding to the coefficients exceeding 0.85 in the calculation is used to calculate the average active power of each child node; the set of child nodes whose average active power is greater than the mean of the average active power of all child nodes is retained. To determine the child node, ; Update the parent-child node data, that is, the remaining active power data at the current parent node Subtract the child node set The active power algebraic sum of the active power matrix composed of the current intelligent circuit breaker sub-nodes Remove the child node set The data sequence represented; Step S4-A5: If the coefficient vector If the number of values exceeding 0.85 does not exceed 1, then execute step S4-A4 and then execute step S4-A6; Otherwise, the number of stacking t Do incremental operation , repeat steps S4-A3 to S4-A4; Step S4-A6: If the level data parameter b =1, then execute step S4-A7; Otherwise, for the number of parent nodes parameter j Do incremental operation , the current parent node remaining active power data Updated to j Repeat steps S4-A2 to S4-A5 to sort out the data of the parent node. j The topology of each parent node is sorted out until all parent nodes of the current level are sorted out; Step S4-A7: Level data parameters b Do incremental operation , reset the number of parent nodes at the current level j =1, child node set , record the child nodes of the current level as the parent nodes of the next level, repeat steps S4-A2 to S4-A6 until all intelligent circuit breakers are divided, and output the topological relationship between the substation assessment summary table and the intelligent circuit breakers at all levels .

7. The self-supervised low-voltage station area topology identification method based on stacked ridge regression according to claim 1 is characterized in that: In step S5: if the topology result is consistent with the existing archival information of the substation, then the user table corresponding to the regression coefficient exceeding the preset value in the hierarchical node where the line loss rate exceeds the warning value is recorded, and it is reported as an abnormal event; otherwise, the line loss rate of each level with the intelligent circuit breaker in the substation measuring equipment as the parent node is calculated, and the result with the lowest average line loss rate is reported.

8. The self-supervised low-voltage substation topology identification method based on stacked ridge regression according to claim 7 is characterized in that: The warning value is 70% and the preset value is 1.5.

Images