Low-voltage distribution area inductive computing integrated topology mapping method, system and medium

Abstract

This application relates to a method, system, and medium for integrated inductive and computational topology mapping of low-voltage distribution areas. The method includes: synchronously acquiring distribution area operation data from an electricity consumption information collection system, a distribution automation system, and a marketing business system, and completing time alignment and standardization processing; constructing a vertical and horizontal dual-dimensional feature fusion system based on the preprocessed data; employing a three-level progressive inference engine to achieve accurate mapping from the feature matrix to the physical topology: the first level is preliminary clustering via machine learning, the second level performs pattern matching for topology refinement, and the third level implements topology inference optimization; based on the optimized topology structure, a coordinate layout is generated, and a standard primitive library is used to automatically generate SVG / DXF format single-line diagrams, establishing an incremental update mechanism to achieve synchronization from the same source and ensure real-time accuracy of the diagram model. This invention can comprehensively utilize dynamic electrical features, automatically infer, and generate standard topology diagrams.

Description

Technical Field

[0001] This application relates to the field of smart distribution network technology, specifically to a method, system, and medium for automatic identification and mapping of transformer area topology using the dynamic propagation law of load characteristics. Background Technology

[0002] Currently, the topology of distribution network areas forms the basis for power grid planning, operation, fault location, and line loss analysis. Traditional methods for obtaining distribution network topology maps mainly rely on manual on-site surveys, digitized drawings, and identification techniques based on single steady-state characteristics. Manual surveys are inefficient, costly, and the drawings are outdated, making them difficult to adapt to frequently changing power grid structures. Methods based on single characteristics such as voltage correlation have limited accuracy in scenarios with similar loads and short electrical distances, and typically only identify "household-transformer" relationships, failing to fully construct multi-level topologies including branches and switches.

[0003] Existing technologies struggle to achieve fully automated, high-precision, and dynamic updates of distribution network topologies. Especially in the context of new power systems, with the large-scale integration of flexible loads such as distributed generation and electric vehicles, the operating states of distribution networks are complex and constantly changing. Static and incomplete topology information has become a bottleneck restricting the lean management and intelligent application of distribution networks. Therefore, there is an urgent need for a method that can comprehensively utilize dynamic electrical characteristics, automatically reason, and generate standard topology maps. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, and medium for integrated inductive and computational topology mapping of low-voltage distribution areas, which can comprehensively utilize dynamic electrical characteristics, automatically reason, and generate standard topology diagrams.

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] In a first aspect, embodiments of this application provide a method for integrated inductive and computational topology mapping of low-voltage distribution areas, comprising the following steps:

[0007] Data acquisition and preprocessing involves synchronously acquiring transformer area operation data from the electricity information acquisition system, distribution automation system, and marketing business system; synchronously capturing two types of key events—load mutation events and switch position change events—as auxiliary criteria; verifying data validity and handling outliers; and completing time alignment and standardization.

[0008] The dual-feature fusion mechanism is constructed based on preprocessed data, and a vertical and horizontal dual-dimensional feature fusion system is built. The vertical dimension focuses on the strong synchronous transmission characteristics of the load, while the horizontal dimension extracts the weak voltage correlation characteristics. Cluster analysis of the load curve shape is used to intelligently determine the type of transformer area, dynamically allocate fusion weights, generate a normalized fusion matrix, and characterize the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning.

[0009] Generative topology mapping is constructed using a three-level progressive inference engine to achieve accurate mapping from feature matrix to physical topology. The first level is machine learning for initial clustering, the second level is pattern matching for topology refinement, and the third level is topology inference optimization.

[0010] The GIS system integrates mapping and dynamic updates. Based on the optimized topology, it generates a coordinate layout, calls the standard graphic element library to automatically generate SVG / DXF format single-line maps, establishes an incremental update mechanism, achieves synchronization from the same source, and ensures that the map model is accurate in real time.

[0011] The transformer area operation data includes three-phase active power, reactive power and three-phase voltage RMS values. The sampling frequency is set to 15 minutes / point, and 672 points of time series are formed by continuous collection for 7 natural days.

[0012] The load mutation event is detected in real time using an improved CUSUM algorithm. A dynamic baseline μ0 is set, and the average load value of the previous 24-hour sliding window is taken to calculate the cumulative deviation. ,

[0013] Where the offset , The baseline window standard deviation, control limits ;when or Time markers are used as candidate mutation points, and transient validation is superimposed: ,

[0014] Residential users must simultaneously meet the following requirements And the rate of change Industrial and commercial users need to meet the following requirements And the rate of change ; Perform temporal clustering on the detected mutation points, at intervals Minutes are merged into a single event;

[0015] Switch change events: Directly extract SOE records from the distribution automation system to obtain the switch state transition timestamp, device ID, and change type, and establish [the necessary information] during the preprocessing stage. Minute-time window matching mechanism: If the time difference between the load change point and the switch change point... If the time interval is less than a second, it is marked as a switch operation-related event. This point will be exempted in subsequent anomaly detection, and an event-related log will be generated for topology reasoning reference.

[0016] In the outlier handling stage, a sliding window approach is adopted for improvement. Criterion: Calculate the local mean using a 12-hour dynamic window. with standard deviation When data points deviate from the window mean by more than The time markers were marked as suspected anomalies; then a secondary trend verification was performed, analyzing the rate of change of the linear regression slope of the 5 data points before and after the anomaly. If there was no corresponding event record to support it and the slope change exceeded 30%, it was confirmed as an anomaly, replaced with the window midpoint, and an audit log was generated. A tiered completion strategy was implemented for missing data: for consecutive missing points ≤ 4, cubic spline interpolation was used to maintain curve smoothness; for missing points 5–24, a seasonal ARIMA model was used to predict and complete the data based on daily load cycle characteristics. Specifically, the effective data of the 24 hours before and after the missing segment were used as the training window, and s=96 was confirmed through spectral analysis. Conventional and seasonal differencing was performed to make the sequence stable, and the ACF / PACF plot and AIC criterion were used to determine the result. The model order is determined by introducing the load of the previous day at the same time, the moving average of the previous hour, and the weekday identifier as exogenous variables to construct a SARIMAX model. This model is used to predict the load value for the missing period and takes the middle part of the extended prediction segment to eliminate boundary effects. If the prediction residual diagnosis fails, it is backed up to linear interpolation and marked as low confidence completion. If the cumulative invalid data ratio of a single table exceeds 15%, it is removed entirely. Finally, time alignment and standardization are completed: the time axis of all terminals in the network is unified by linear interpolation based on the TTU timestamp; the load data is scaled to the [0,1] interval using Min-Max; the voltage data is standardized by Z-score to eliminate the difference in units and improve the numerical stability of subsequent feature extraction, and a structured dataset is output.

[0017] Vertical dimension focuses on the characteristics of strong synchronous load transmission: an improved CUSUM algorithm is used to accurately detect load step changes, and adjacent change points with a time interval of less than 2 minutes are clustered into the same event; for each valid event, the response time difference between devices is calculated. The waveform similarity was calculated using a dynamic time warping algorithm based on the load curves 30 minutes before and after the mutation. This leads to the construction of the load transmission strength matrix. Weak voltage correlation features were extracted in the horizontal dimension: the Pearson correlation coefficient between meters was calculated by sliding the data with a 24-hour window and a 4-hour step size. Electrical distance is constructed by averaging multiple windows. The voltage correlation strength was obtained by normalization. To achieve adaptive fusion, the load curve shape is first analyzed using K-means++ clustering to intelligently determine the type of transformer area and dynamically allocate fusion weights: 0.35 for residential-dominated types and 0.65 for voltage correlation types; 0.75 and 0.25 for industrial and commercial-dominated types; and 0.55 and 0.45 for mixed types. Finally, a normalized fusion matrix is ​​generated to characterize the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning.

[0018] The first stage is preliminary clustering using machine learning: merging the matrix... As the similarity input, construct the Laplacian matrix. Fiedler vectors are extracted for spectral clustering; the K value is adaptively determined using silhouette coefficients, and preliminary device grouping is output. The second level of execution involves topology refinement through pattern matching: It calls the built-in library of six typical transformer area topology templates, calculates the graph edit distance between the clustering results and each template, optimizes the matching path using the A* search algorithm, and selects the matching score. The template serves as the basic topology skeleton, and the third level implements topology reasoning optimization: strictly embedding electrical physical constraints for closed-loop verification—load balance verification requires the relative error of node input and output power. Voltage drop verification comparison with actual measurement The error between the theoretical calculation value and the actual value is ≤10%; the connectivity check ensures that there are no isolated nodes and the branch level is ≤4 and the number of devices in a single branch is ≤30; for the connection relationship that fails the check, the device affiliation is adjusted back according to the transmission strength matrix, and iterative optimization is performed until all constraints are met.

[0019] Secondly, embodiments of this application provide a low-voltage distribution area inductive computing integrated topology mapping system, including,

[0020] The data acquisition and preprocessing module synchronously acquires transformer area operation data from the electricity information acquisition system, distribution automation system and marketing business system, synchronously captures two types of key events, load change events and switch position change events, as auxiliary criteria, verifies data validity and handles outliers, and completes time alignment and standardization.

[0021] The dual-feature fusion mechanism construction module builds a vertical and horizontal dual-dimensional feature fusion system based on preprocessed data. The vertical dimension focuses on the strong synchronous transmission characteristics of the load, while the horizontal dimension extracts the weak voltage correlation characteristics. Cluster analysis of the load curve shape intelligently determines the type of transformer area, dynamically allocates fusion weights, generates a normalized fusion matrix, and characterizes the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning.

[0022] The generative topology mapping construction module uses a three-level progressive inference engine to achieve accurate mapping from feature matrix to physical topology. The first level is machine learning for preliminary clustering, the second level performs pattern matching for topology refinement, and the third level implements topology inference optimization.

[0023] The GIS fusion mapping and dynamic update module generates a coordinate layout based on the optimized topology, automatically generates SVG / DXF format single-line maps by calling the standard graphic element library, establishes an incremental update mechanism, achieves synchronization from the same source, and ensures that the map model is accurate in real time.

[0024] Thirdly, embodiments of this application provide a computer-readable storage medium, characterized in that the computer-readable storage medium stores program code, which, when executed by a processor, implements the steps of a low-voltage distribution area inductive computing integrated topology mapping method as described above.

[0025] Compared with the prior art, the beneficial effects of the present invention are:

[0026] This method can be used to quickly and accurately identify the topology of transformer substations using data analytics, helping power companies to automatically update their records. It eliminates the need for manual on-site inspections, rapidly locating issues such as incorrect transformer-household relationships and phase errors, significantly improving operational efficiency.

[0027] It can automatically generate single-line diagrams of transformer substations that conform to the State Grid standard, solving the problem of "data and diagram separation" and realizing the same source maintenance of data and graphics.

[0028] Based on the dynamic propagation law of load switching and steady-state operation characteristics, it can accurately distinguish vertical hierarchical relationships and horizontal branch relationships, with an identification accuracy rate higher than that of traditional single feature analysis methods. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart illustrating the specific process of the method in an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. It should be noted that similar reference numerals and letters in the following drawings indicate similar items; therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0032] The terms “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0033] The terms “first,” “second,” etc., are used only to distinguish one entity or operation from another, and should not be construed as indicating or implying relative importance, nor as requiring or implying any such actual relationship or order between these entities or operations.

[0034] The specific implementation process of this invention is as follows: Figure 1 The steps are as follows:

[0035] Step 1: Data Acquisition and Preprocessing

[0036] This step synchronously acquires transformer area operation data from the electricity consumption information collection system, distribution automation system, and marketing business system. Core acquisition parameters include three-phase active power (kW), reactive power (kvar), and three-phase voltage RMS value (V). The sampling frequency is strictly set to 15 minutes / point, and continuous collection for 7 calendar days forms a 672-point time series. Two types of key events are simultaneously captured as auxiliary criteria:

[0037] (1) Load mutation events: The improved CUSUM algorithm is used for real-time detection. A dynamic baseline μ0 is set (the average load of the previous 24-hour sliding window), and the cumulative deviation is calculated:

[0038] Where the offset ( (baseline window standard deviation), control limits ;when or Time markers are used as candidate mutation points, and transient validation is superimposed:

[0039] Residential users must simultaneously meet the following requirements And the rate of change Industrial and commercial users need to meet the following requirements And the rate of change ; Perform temporal clustering on the detected mutation points, at intervals Minutes are merged into a single event.

[0040] Switch change events: Directly extract the SOE (Sequence of Events) records from the distribution automation system to obtain the switch state transition timestamp, device ID, and change type (closed / open); establish during the preprocessing stage. Minute-time window matching mechanism: If the time difference between the load change point and the switch change point... If the time interval is less than 1 second, it is marked as a "switch operation associated event". This point will be exempted in subsequent anomaly determination, and an event association log will be generated for topology reasoning reference.

[0041] Data validity verification requires that at least 85% of the data points in a single table are valid; those that do not meet this requirement are automatically marked and removed by the terminal. In the outlier handling process, a sliding window approach is used for improvement. Criterion: Calculate the local mean using a 12-hour (48 sampling points) dynamic window. with standard deviation When data points deviate from the window mean by more than The time marker is marked as a suspected anomaly; then a secondary trend verification is performed, analyzing the rate of change of the linear regression slope of the 5 data points before and after the anomaly. If there is no corresponding event record to support it and the slope change exceeds 30%, it is confirmed as an anomaly, replaced with the window midpoint, and an audit log is generated. A tiered completion strategy is implemented for missing data: for consecutive missing data of ≤4 points (≤1 hour), cubic spline interpolation is used to maintain curve smoothness; for missing data of 5–24 points (1–6 hours), the seasonal ARIMA model is used in conjunction with daily load cycle characteristics for prediction and completion. Specifically, the effective data of 24 hours before and after the missing segment are used as the training window, and s=96 (24-hour cycle) is confirmed through spectrum analysis. Conventional and seasonal differencing is performed to make the sequence stable, and the ACF / PACF plot and AIC criterion are combined to determine the time frame. The model order was determined by introducing the load of the previous day at the same time, the moving average of the previous hour, and the weekday identifier as exogenous variables to construct a SARIMAX model. This model was used to predict the load values ​​for the missing period and the middle part of the extended prediction segment was taken to eliminate boundary effects. If the prediction residual diagnosis failed, it was reverted to linear interpolation and marked as low-confidence completion. Data with more than 15% of the data in a single table were removed. Finally, time alignment and standardization were completed: the time axis of all terminals across the network was unified by linear interpolation based on the TTU timestamp; load data was scaled to the [0,1] interval using Min-Max; voltage data was Z-score standardized to eliminate dimensional differences and improve the numerical stability of subsequent feature extraction. A structured dataset was output for use in step 2.

[0042] Step 2: Construction of the dual-feature fusion mechanism

[0043] Based on the preprocessed data, this step constructs a dual-dimensional feature fusion system encompassing both vertical and horizontal dimensions. The vertical dimension focuses on the strong synchronous transmission characteristics of the load: an improved CUSUM algorithm is used to accurately detect load step abrupt changes (parameter settings: reference mean μ0 is taken as the baseline load, offset k=0.5σ, control limit h=5σ), clustering adjacent abrupt change points with time intervals less than 2 minutes into the same event; for each valid event, the response time difference between devices is calculated. The waveform similarity was calculated using the Dynamic Time Warping (DTW) algorithm based on the load curves 30 minutes before and after the mutation. This leads to the construction of the load transmission strength matrix. Horizontal dimension extraction of weak voltage correlation features: Pearson correlation coefficient between meters is calculated using a 24-hour window and a 4-hour step size. Electrical distance is constructed by averaging multiple windows. The voltage correlation strength was obtained by normalization. To achieve adaptive fusion, the load curve morphology is first analyzed using K-means++ clustering to intelligently determine the transformer substation type, and fusion weights are dynamically allocated: 0.35 for residential-dominated substations and 0.65 for voltage correlation; 0.75 and 0.25 for industrial and commercial-dominated substations; and 0.55 and 0.45 for mixed substations. Finally, a normalized fusion matrix is ​​generated to characterize the coupling relationship between vertical transmission and lateral branches between devices, providing high-confidence input for topology inference.

[0044] Step 3: Construction of Generative Topology Maps

[0045] This step employs a three-level progressive inference engine to achieve a precise mapping from the feature matrix to the physical topology. The first level is preliminary clustering via machine learning: merging the feature matrix... As the similarity input, construct the Laplacian matrix. Fiedler vectors are extracted for spectral clustering; the K value is adaptively determined using silhouette coefficients, and preliminary device grouping is output. The second level of execution involves pattern matching and topology refinement: It calls the built-in library of six typical transformer area topology templates, calculates the graph edit distance (GED) between the clustering results and each template, optimizes the matching path using the A* search algorithm, and selects the matching score. The template serves as the basic topology skeleton. The third level implements topology reasoning optimization: strictly embedding electrical physics constraints for closed-loop verification—load balance verification requires the relative error of node input and output power. Voltage drop verification comparison with actual measurement The error between the theoretical calculation value and the actual value is ≤10%; connectivity verification ensures no isolated nodes and that the branch level is ≤4 and the number of devices per branch is ≤30; for connections that fail verification, the device affiliation is adjusted backtrackingly based on the transmission strength matrix, and iterative optimization is performed until all constraints are met. This three-level reasoning mechanism forms a closed-loop logic of "data-driven clustering → knowledge-guided matching → physical rule correction", which significantly improves the engineering credibility and on-site consistency of the topology.

[0046] Step 4: GIS fusion into maps and dynamic updates

[0047] Based on the optimized topology, an improved Fruchterman-Reingold force-directed algorithm is used to generate the coordinate layout. The Q / GDW 1799-2022 standard element library is used to automatically generate SVG / DXF format single-line diagrams. An incremental update mechanism is established so that only local subtrees are redrawn when the equipment change ratio is less than 15%. The system is synchronized with the PMS / GIS system through the IEC 61968 interface to ensure that the map model is accurate in real time.

[0048] This application provides a low-voltage distribution area inductive computing integrated topology mapping system, including,

[0049] The data acquisition and preprocessing module synchronously acquires transformer area operation data from the electricity information acquisition system, distribution automation system and marketing business system, synchronously captures two types of key events, load change events and switch position change events, as auxiliary criteria, verifies data validity and handles outliers, and completes time alignment and standardization.

[0050] The dual-feature fusion mechanism construction module builds a vertical and horizontal dual-dimensional feature fusion system based on preprocessed data. The vertical dimension focuses on the strong synchronous transmission characteristics of the load, while the horizontal dimension extracts the weak voltage correlation characteristics. Cluster analysis of the load curve shape intelligently determines the type of transformer area, dynamically allocates fusion weights, generates a normalized fusion matrix, and characterizes the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning.

[0051] The generative topology mapping construction module uses a three-level progressive inference engine to achieve accurate mapping from feature matrix to physical topology. The first level is machine learning for preliminary clustering, the second level performs pattern matching for topology refinement, and the third level implements topology inference optimization.

[0052] The GIS fusion mapping and dynamic update module generates a coordinate layout based on the optimized topology, automatically generates SVG / DXF format single-line maps by calling the standard graphic element library, establishes an incremental update mechanism, achieves synchronization from the same source, and ensures that the map model is accurate in real time.

[0053] This application provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores program code, which, when executed by a processor, implements the steps of a low-voltage distribution area inductive computing integrated topology mapping method as described above.

[0054] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0055] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0056] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0057] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0058] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0059] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0060] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0061] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for integrated inductive computation topology mapping of low-voltage distribution areas, characterized in that, Includes the following steps: Data acquisition and preprocessing involves synchronously acquiring transformer area operation data from the electricity information acquisition system, distribution automation system, and marketing business system; synchronously capturing two types of key events—load mutation events and switch position change events—as auxiliary criteria; verifying data validity and handling outliers; and completing time alignment and standardization. The dual-feature fusion mechanism is constructed based on preprocessed data, and a vertical and horizontal dual-dimensional feature fusion system is built. The vertical dimension focuses on the strong synchronous transmission characteristics of the load, while the horizontal dimension extracts the weak voltage correlation characteristics. Cluster analysis of the load curve shape is used to intelligently determine the type of transformer area, dynamically allocate fusion weights, generate a normalized fusion matrix, and characterize the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning. Generative topology mapping is constructed using a three-level progressive inference engine to achieve accurate mapping from feature matrix to physical topology. The first level is machine learning for initial clustering, the second level is pattern matching for topology refinement, and the third level is topology inference optimization. The GIS system integrates mapping and dynamic updates. Based on the optimized topology, it generates a coordinate layout, calls the standard graphic element library to automatically generate SVG / DXF format single-line maps, establishes an incremental update mechanism, achieves synchronization from the same source, and ensures that the map model is accurate in real time.

2. The method for integrated inductive computing topology mapping of a low-voltage distribution area according to claim 1, characterized in that, The transformer area operation data includes three-phase active power, reactive power and three-phase voltage RMS values. The sampling frequency is set to 15 minutes / point, and 672 points of time series are formed by continuous collection for 7 natural days.

3. The method for integrated inductive computing topology mapping of a low-voltage distribution area according to claim 1, characterized in that, The load mutation event is detected in real time using an improved CUSUM algorithm. A dynamic baseline μ0 is set, and the average load value of the previous 24-hour sliding window is taken to calculate the cumulative deviation. , , Where the offset , The baseline window standard deviation, control limits ;when or Time markers are used as candidate mutation points, and transient validation is superimposed: , Residential users must simultaneously meet the following requirements And the rate of change Industrial and commercial users need to meet the following requirements And the rate of change ; Perform temporal clustering on the detected mutation points, at intervals Minutes are merged into a single event; Switch change events: Directly extract SOE records from the distribution automation system to obtain the switch state transition timestamp, device ID, and change type, and establish [the necessary information] during the preprocessing stage. Minute-time window matching mechanism: If the time difference between the load change point and the switch change point... If the time interval is less than a second, it is marked as a switch operation-related event. This point will be exempted in subsequent anomaly detection, and an event-related log will be generated for topology reasoning reference.

4. The method for integrated inductive computing topology mapping of low-voltage distribution areas according to claim 1, characterized in that, In the outlier handling stage, a sliding window approach is adopted for improvement. Criterion: Calculate the local mean using a 12-hour dynamic window. with standard deviation When data points deviate from the window mean by more than The time was marked as a suspected anomaly; Next, a secondary trend verification was performed, analyzing the rate of change of the linear regression slope of the 5 data points before and after the outlier. If there was no corresponding event record to support the outlier and the slope change exceeded 30%, it was confirmed as an outlier, replaced with the window midpoint, and an audit log was generated. A tiered completion strategy was implemented for missing data: for consecutive missing data of ≤4 points, cubic spline interpolation was used to maintain curve smoothness; for missing data of 5–24 points, a seasonal ARIMA model was used to predict and complete the data based on daily load cycle characteristics. Specifically, the effective data of the 24 hours before and after the missing segment were used as the training window, and spectral analysis confirmed s=96. Conventional and seasonal differencing was performed to make the sequence stable, and the ACF / PACF plot and AIC criterion were used to determine the... The model order is determined by introducing the load of the previous day at the same time, the moving average of the previous hour, and the weekday identifier as exogenous variables to construct a SARIMAX model. This model is used to predict the load value for the missing period and takes the middle part of the extended prediction segment to eliminate boundary effects. If the prediction residual diagnosis fails, it is backed up to linear interpolation and marked as low confidence completion. If the cumulative invalid data ratio of a single table exceeds 15%, it is removed entirely. Finally, time alignment and standardization are completed: the time axis of all terminals in the network is unified by linear interpolation based on the TTU timestamp; the load data is scaled to the [0,1] interval using Min-Max; the voltage data is standardized by Z-score to eliminate the difference in units and improve the numerical stability of subsequent feature extraction, and a structured dataset is output.

5. The method for integrated inductive calculation and topology mapping of low-voltage distribution areas according to claim 1, characterized in that, Vertical dimension focuses on the characteristics of strong synchronous load transmission: an improved CUSUM algorithm is used to accurately detect load step changes, and adjacent change points with a time interval of less than 2 minutes are clustered into the same event; for each valid event, the response time difference between devices is calculated. The waveform similarity was calculated using a dynamic time warping algorithm based on the load curves 30 minutes before and after the mutation. This leads to the construction of the load transmission strength matrix. Weak voltage correlation features were extracted in the horizontal dimension: the Pearson correlation coefficient between meters was calculated by sliding the data with a 24-hour window and a 4-hour step size. Electrical distance is constructed by averaging multiple windows. The voltage correlation strength was obtained by normalization. To achieve adaptive fusion, the load curve shape is first analyzed using K-means++ clustering to intelligently determine the type of transformer area and dynamically allocate fusion weights: 0.35 for residential-dominated types and 0.65 for voltage correlation types; 0.75 and 0.25 for industrial and commercial-dominated types; and 0.55 and 0.45 for mixed types. Finally, a normalized fusion matrix is ​​generated to characterize the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning.

6. The method for integrated inductive computing topology mapping of low-voltage distribution areas according to claim 1, characterized in that, The first stage is preliminary clustering using machine learning: merging the matrix... As the similarity input, construct the Laplacian matrix. Fiedler vectors are extracted for spectral clustering; the K value is adaptively determined using silhouette coefficients, and preliminary device grouping is output. The second level of execution involves topology refinement through pattern matching: It calls the built-in library of six typical transformer area topology templates, calculates the graph edit distance between the clustering results and each template, optimizes the matching path using the A* search algorithm, and selects the matching score. The template serves as the basic topology skeleton, and the third level implements topology reasoning optimization: strictly embedding electrical physical constraints for closed-loop verification—load balance verification requires the relative error of node input and output power. ; Voltage drop verification comparison actual measurement The error between the theoretical calculation value and the actual value is ≤10%; the connectivity check ensures that there are no isolated nodes and the branch level is ≤4 and the number of devices in a single branch is ≤30; for the connection relationship that fails the check, the device affiliation is adjusted back according to the transmission strength matrix, and iterative optimization is performed until all constraints are met.

7. A low-voltage distribution area inductive computing integrated topology mapping system, characterized in that, include, The data acquisition and preprocessing module synchronously acquires transformer area operation data from the electricity information acquisition system, distribution automation system and marketing business system, synchronously captures two types of key events, load change events and switch position change events, as auxiliary criteria, verifies data validity and handles outliers, and completes time alignment and standardization. The dual-feature fusion mechanism construction module builds a vertical and horizontal dual-dimensional feature fusion system based on preprocessed data. The vertical dimension focuses on the strong synchronous transmission characteristics of the load, while the horizontal dimension extracts the weak voltage correlation characteristics. Cluster analysis of the load curve shape intelligently determines the type of transformer area, dynamically allocates fusion weights, generates a normalized fusion matrix, and characterizes the coupling relationship between vertical transmission and horizontal branches between devices, providing high-confidence input for topology reasoning. The generative topology mapping construction module uses a three-level progressive inference engine to achieve accurate mapping from feature matrix to physical topology. The first level is machine learning for preliminary clustering, the second level performs pattern matching for topology refinement, and the third level implements topology inference optimization. The GIS fusion mapping and dynamic update module generates a coordinate layout based on the optimized topology, automatically generates SVG / DXF format single-line maps by calling the standard graphic element library, establishes an incremental update mechanism, achieves synchronization from the same source, and ensures that the map model is accurate in real time.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program code, which, when executed by a processor, implements the steps of a low-voltage distribution area inductive computing integrated topology mapping method as described in any one of claims 1 to 6.

Images