Low-voltage distribution area topology identification method and system for load switching event detection

Abstract

The present application belongs to the technical field of low-voltage power distribution automation, and more particularly to a low-voltage power distribution area topology identification method and system for load switching event detection. The present application comprises the following steps: S1, preparation for area load switching event detection; S2, active power time sequence construction; S3, load switching transient event detection; S4, event parameter standardization calculation; S5, cross-level homologous event matching; S6, area topology automatic construction and correction. The present application constructs a four-level monitoring node architecture through the inventory HPLC communication network, completes the millisecond-level time synchronization of all network nodes, calculates the active power by integral calculation of the whole cycle, extracts the load switching transient mutation characteristics by DB5 orthogonal wavelet four-layer tower decomposition, matches the cross-level homologous events based on the three conditions of event type-time interval-amplitude deviation, and corrects by secondary retest, thereby realizing the automatic identification effect of the four-level tree topology of the area.

Description

Technical Field

[0001] This invention belongs to the field of low-voltage power distribution automation technology, and more specifically, relates to a method and system for identifying the topology of low-voltage power distribution substations for load switching event detection. Background Technology

[0002] Low-voltage distribution transformer areas are the smallest basic management unit for distribution network operation and maintenance, generally exhibiting a four-level tree-like topology structure of "transformer-distribution box-meter box-meter". Accurate transformer area topology relationships are a fundamental prerequisite for achieving line loss management, fault location, three-phase load optimization, and refined distribution network operation and maintenance. Currently, low-voltage transformer area topology identification technologies mainly include three categories: manual mapping and archiving, active signal injection identification, and data-driven identification based on steady-state time-series data correlation. Manual mapping relies on maintenance personnel manually verifying line wiring relationships on-site to create topology files, which is labor-intensive and time-consuming. Furthermore, low-voltage transformer areas experience frequent changes such as user capacity increases / decreases, line upgrades, and new photovoltaic installations, making real-time updates of manual files impossible, resulting in long-term problems of topology inconsistencies and inaccurate user-transformer relationships. Active signal injection identification requires additional signal generation and injection modules at the meter, meter box, and branch box terminals, necessitating large-scale modifications to existing terminal equipment and resulting in significant engineering investment. Furthermore, the injected characteristic signals are prone to harmonic superposition with power electronic devices such as photovoltaic inverters and charging piles, causing power quality disturbances. In high-noise and high-disturbance areas, signal submersion and identification failure are likely to occur. Steady-state time-series correlation identification only judges the node correlation based on the similarity of steady-state voltage and load waveforms. It can only achieve in-phase and in-area clustering, and cannot use electrical transmission laws to distinguish upstream and downstream subordinate relationships. It is difficult to construct a complete four-level tree topology. In areas with high penetration of distributed photovoltaic and charging piles, photovoltaic reverse output and random load fluctuations will seriously destroy the steady-state waveform correlation, resulting in a large number of false correlations across branches and failure of correlations within branches.

[0003] Patent document CN116436152A discloses an intelligent low-voltage distribution substation topology identification method based on feature information correlation. This method constructs a low-voltage distribution substation topology identification architecture system based on an intelligent fusion terminal and high-speed power line carrier communication technology. It extracts and analyzes feature datasets collected by intelligent monitoring units at each branch box, smart meter box, and user side, calculates the correlation between the feature information of each branch incoming and outgoing line within the substation, and performs comprehensive topology analysis by combining the signal-to-noise ratio and zero-crossing phase offset of each node within the substation. Although this achieves dynamic management of low-voltage substation topology identification, it still relies on steady-state feature datasets and signal-to-noise ratio, zero-crossing phase offset, and other signal features for correlation analysis. In substations with high penetration of new energy, the steady-state waveform correlation is easily disrupted by photovoltaic reverse output and random load fluctuations, making it impossible to accurately distinguish upstream and downstream subordinate levels using the transient transmission law of load disturbances.

[0004] Patent document CN117277567A discloses a method for identifying the topology of a low-voltage distribution transformer area. This method uses an intelligent fusion terminal and an intelligent detection unit to network the communication nodes of the corresponding distribution transformer area. Power line carrier signals are loaded into the distribution transformer, and the zero-crossing times, power line carrier communication energy, and noise energy of each communication node are collected. The zero-crossing phase difference and carrier signal-to-noise ratio are calculated to determine whether each communication node belongs to the same branch and generate a transformer area topology connection table. Although a total physical connection table of the transformer area is generated through hierarchical logical judgment, it still suffers from problems such as reliance on carrier signal characteristics and similarity analysis, the need for active excitation by loading power line carrier signals on the transformer side, unutilized natural disturbance resources such as load switching, and the inability to achieve cross-level homogeneous matching and automatic determination of hierarchical affiliation based on user-side disturbance events.

[0005] Patent document CN112234605A discloses a method for identifying the relationship between a user and a transformer based on load characteristics of electrical appliance start-stop. It utilizes the power disturbance generated by the natural start-stop of user-side electrical appliances (without actively injected signals) as an excitation source. Between the sub-meters and the main meter, it matches three features: event trigger time, power amplitude, and direction of change, to identify the relationship between the user and the transformer. Although it can achieve single-level user-transformer relationship identification using natural disturbances without actively injecting signals, it only achieves a single-level user-transformer relationship determination between the sub-meters and the main meter, and does not constitute a comprehensive method. The four-level cross-layer topology identification architecture of "transformer-distribution box-meter box-meter" was constructed, but it did not adopt the whole-cycle discrete integral calculation of active power and the DB5 orthogonal wavelet four-layer tower decomposition to extract the transient change features of load switching. Furthermore, it did not provide a bottom-up step-by-step splicing mechanism for the four-level topology based on millisecond-level time synchronization, cross-level homogeneous event step-by-step matching, three-level high-frequency detail double-layer threshold joint judgment, unique membership judgment, and secondary retest closed-loop correction. Therefore, it was impossible to completely and accurately restore the four-level three-dimensional topology structure in active transformer areas with high penetration of new energy and drastic load fluctuations.

[0006] In summary, there is an urgent need to provide a method and system for identifying the topology of low-voltage distribution transformer areas by detecting load switching events, in order to solve the problems of high cost of upgrading low-voltage transformer area topology identification, large power quality interference, poor adaptability to new energy scenarios, inability to distinguish topology levels, and inability to dynamically update the topology in the existing technology. Summary of the Invention

[0007] The technical problem to be solved by this invention is: to construct a four-level monitoring node architecture of "transformer-distribution box-meter box-meter" through the existing HPLC communication network and complete millisecond-level time synchronization of all network nodes; to calculate active power with integer-cycle discrete integral and extract the transient change features of load switching using DB5 orthogonal wavelet four-layer tower decomposition; and to perform cross-level homogeneous event matching based on event type-time interval-amplitude deviation triple conditions and supplemented by secondary retesting and correction, so as to solve the problems mentioned in the background technology, such as high cost of low-voltage distribution area topology identification and transformation, large power quality interference, poor adaptability to new energy scenarios, inability to distinguish topology levels, and inability to dynamically update the topology, and to achieve low-cost, interference-free, hierarchical distinguishable, and dynamically updateable automatic identification effect of four-level tree topology of distribution areas.

[0008] The technical terms related to this invention are explained as follows: HPLC: High-speed power line broadband carrier communication technology, which enables broadband communication and time synchronization of all network nodes through power lines.

[0009] Master node: A monitoring node located on the low-voltage side of the distribution transformer, responsible for sending detection signals, summarizing event parameters, and performing topology identification.

[0010] Sub-nodes: Monitoring nodes set up at distribution boxes, meter boxes, and electricity meters are responsible for collecting electrical quantities at their respective locations and reporting event parameters.

[0011] Line load switching events: Power step disturbance events caused by the connection or disconnection of user electrical equipment.

[0012] Active power time series: Active power data sequence of each monitoring node arranged in chronological order, retaining transient fluctuation characteristics at the moment of load switching.

[0013] DB5 Orthogonal Wavelet: A 5th-order Daubechies orthogonal wavelet used for extracting transient change features in active power time series.

[0014] Tower decomposition: In wavelet multiresolution analysis, the process of iteratively decomposing low-frequency approximate components into low-frequency approximate components and high-frequency detail components layer by layer.

[0015] Sudden power amplitude: The change in steady-state baseline power caused by a load switching event, characterizing the intensity of load fluctuations during a single disturbance.

[0016] Same-source event matching: The process of determining whether load switching events reported by nodes at different levels originate from the same load disturbance event.

[0017] Subordination identification: The process of determining the upstream and downstream subordination relationships of nodes at each level of the meter, meter box, and distribution box based on the matching results of events from the same source.

[0018] The technical problem to be solved by this invention is achieved by the following technical solution: A method for identifying the topology of a low-voltage distribution transformer area based on load switching event detection includes the following steps: S1. Preparation for load switching event detection in the distribution area: Through the HPLC communication network already deployed in the low-voltage distribution area, high-frequency synchronous acquisition of voltage and current is carried out on the monitoring nodes at all levels of distribution transformers, distribution boxes, meter boxes, and meters in the distribution area, and the time synchronization of all network nodes is completed through the network time synchronization mechanism of the HPLC communication network. S2. Active power time series construction: Within a specified time period, the sub-nodes calculate the active power using a sliding window method and an integer-cycle discrete integration algorithm based on the synchronously acquired voltage and current time series data, and generate an active power time series. S3. Load switching transient event detection: Discrete wavelet transform is used to extract transient change features from the active power time series to detect line load switching events; S4. Event Parameter Standardization Calculation: Based on the disturbance time located by the transient mutation feature extraction, the steady-state baseline difference method is used to calculate the mutation power amplitude, forming event parameters that include the event occurrence time, event type, and mutation power amplitude, and then reporting them to the master node. S5. Cross-level same-source event matching: The master node matches the line load switching events reported by nodes at each level based on the line load switching event parameters reported by sub-nodes across the entire network; S6. Automatic construction and correction of transformer area topology: Based on the matching results, the affiliation relationship between nodes at each level of meters, meter boxes, and distribution boxes is identified to obtain the transformer area topology structure. Nodes that have not been matched are included in the pending list, and secondary retesting and identification are completed by extending the detection window to correct the topology identification results.

[0019] This technical solution utilizes the existing HPLC communication network to achieve high-frequency synchronous acquisition of voltage and current at four monitoring nodes and millisecond-level time synchronization across the entire network. It uses the natural load switching disturbances generated by the daily start-up and shutdown of electrical equipment in the distribution area as the excitation source for topology identification. Within a specified time period, sub-nodes calculate the active power time series based on the synchronously acquired data and extract transient change characteristics. They detect line load switching events and report the event occurrence time, event type, and change power amplitude. The master node performs cross-level matching based on the event parameters reported across the entire network, layer by layer identifying the affiliation relationships of meters, meter boxes, and distribution boxes, ultimately obtaining the complete distribution area topology. The entire solution fully reuses the HPLC smart meters, branch monitoring terminals, integrated terminals, and existing communication and measurement systems already deployed in the distribution area. It requires no additional dedicated hardware such as signal injection modules or pulse generators, and no on-site construction work such as equipment replacement or line modification. There is no artificial harmonic or pulse signal output, and it will not generate signal superposition or harmonic interference with power electronic equipment such as photovoltaic inverters, charging piles, and frequency conversion loads. At the same time, by utilizing the physical law of load disturbance being transmitted in series along the line and attenuating step by step, it can not only determine the subordinate relationship between nodes and branches, but also accurately distinguish the upstream and downstream parent-child node hierarchy, completely restoring the four-level three-dimensional topology structure of the low-voltage distribution area: "transformer-distribution box-meter box-meter". The integrity and accuracy of topology data are significantly improved.

[0020] The monitoring nodes in the distribution area include main nodes set up on the low-voltage side of the distribution transformer, and sub-nodes set up at distribution boxes, meter boxes, and electricity meters, forming a four-level hierarchical structure from top to bottom: "transformer-distribution box-meter box-electricity meter". Each level of monitoring node performs high-frequency synchronous acquisition of line voltage and current time sequence operation data at its location, and the synchronization accuracy between nodes reaches the millisecond level.

[0021] This technical solution constructs a global synchronous measurement network covering four levels of nodes: transformers, distribution boxes, meter boxes, and meters, through a hierarchical deployment of master and child nodes. Each level of node independently collects voltage and current timing data at its location, and achieves millisecond-level time synchronization across the entire network through the HPLC network time synchronization mechanism. This ensures the consistency of electrical data timing across the entire network and provides an accurate time reference for the matching of load switching events across levels, fundamentally avoiding false event associations caused by misaligned sampling times of each node.

[0022] When a topology identification requirement arises in a transformer area, the master node sends a signal to all sub-nodes in the network through the HPLC communication network, requiring the sub-nodes to perform line load switching event detection within a specified time period and report the detection results to the master node. The specified time period is a pre-set limited duration detection window.

[0023] This technical solution involves the master node actively initiating detection signaling and configuring a preset time window to concentrate the topology identification task within a specified time period. The typical value of the specified time period is one day. This ensures that a sufficient number of load switching event samples are collected within the time window, while avoiding data redundancy and communication overhead caused by long-term continuous collection. This allows the topology identification process to be triggered on demand and executed according to the window, improving system operating efficiency and manageability.

[0024] The child nodes calculate active power based on synchronously acquired discrete voltage and current sampling sequences, using a sliding window approach and an integer-cycle discrete integration algorithm. The active power calculation formula is as follows:

[0025] In the formula: This represents the number of discrete sampling points within a single power frequency cycle. , These are the first and second units within the power frequency cycle. The instantaneous voltage and instantaneous current values ​​at each sampling time are used, and the number of sliding points is the preset window sliding step size.

[0026] This technical solution calculates active power using an integer-cycle discrete integral algorithm. It eliminates instantaneous power fluctuations within the cycle by using the entire power frequency cycle as the integration interval. The calculation is performed frame by frame using a sliding window method to generate a continuous and smooth time series sequence of active power across the entire network. This preserves the transient fluctuation characteristics at the moment of load switching while filtering out measurement noise within the power frequency cycle. It provides input time series data with excellent signal-to-noise ratio for subsequent wavelet multi-scale decomposition, avoiding the transient change characteristics being submerged by noise due to power calculation errors.

[0027] Discrete wavelet transform is used to extract transient change features from the active power time series; DB5 orthogonal wavelet is used to perform a four-level pyramid decomposition of the active power time series, extracting low-frequency approximate components and high-frequency detail components layer by layer. The decomposition iterative calculation formula is as follows:

[0028]

[0029] In the formula: This is the original active power time series. For the first Low-frequency approximation components of the layer, , The first Low-frequency approximate components and high-frequency detail components of the layer. These are the low-pass filter coefficients corresponding to the scaling function. These are the high-pass filter coefficients corresponding to the wavelet function. The number of decomposition layers, This refers to the filter coefficient index.

[0030] This technical solution uses DB5 orthogonal wavelet four-level tower decomposition to separate the active power time series into low-frequency trend components and high-frequency detail components. After four-level decomposition, four levels of high-frequency detail components are obtained, and each layer of components has clear physical scale differences: the shallow high-frequency components correspond to high-frequency measurement noise and small random load fluctuations, while the deep high-frequency components correspond to medium- and low-frequency, longer-lasting, and more stable equipment step switching disturbances, which are the core identification features of effective load switching events. Through multi-scale decomposition, the real load switching disturbances are highlighted from the noise background, significantly improving the robustness and accuracy of transient change feature extraction.

[0031] Thresholds are set for the Level 3 and Level 4 high-frequency detail components. If the Level 3 and Level 4 high-frequency detail components simultaneously exhibit significant peak values ​​exceeding their respective thresholds at the same time point, then that time point is determined to be the actual load disturbance time point. Event classification is completed by combining the original power time series change trend: if the power time series corresponding to the disturbance time point shows a stable step upward trend, it is determined to be a load input event; if the power time series corresponding to the disturbance time point shows a stable step downward trend, it is determined to be a load shedding event.

[0032] This technical solution employs a dual-layer threshold joint judgment mechanism for three- and four-level high-frequency detail components. It requires that the two layers of high-frequency components simultaneously exceed their respective thresholds before a true load disturbance is identified, effectively suppressing false triggering caused by single-scale noise spikes. By combining the step direction of the original power time series, it completes the classification of load input and load shedding events, ensuring that the event detection results include both the time of disturbance occurrence and the type of disturbance, providing complete event attribute information for subsequent cross-level homogeneous matching.

[0033] The amplitude of the sudden change power is calculated using the baseline difference method before and after the event's steady state: Based on the disturbance time located by the transient sudden change feature extraction, a preset steady-state window of data is extracted forward, and a steady-state window of data after the disturbance is completed is extracted backward. The steady-state baseline power before and after the switch is calculated respectively. The formula for calculating the amplitude of the sudden change power is as follows:

[0034]

[0035]

[0036] In the formula: The steady-state baseline power before switching, This represents the steady-state baseline power after switching. For precise wavelet positioning of event moments. , These represent the lengths of the initial and final steady-state windows, respectively; the final mutation power amplitude is... For load input events, meet the requirements. Power increases positively by a step; for load shedding events, it satisfies... The power drops negatively by a step.

[0037] This technical solution calculates the amplitude of sudden power changes using the baseline difference method before and after the steady state of an event. It uses the perturbation moment precisely located by wavelet as a reference to extract steady-state data windows before and after the perturbation and calculate the steady-state average power. The difference between the steady-state baseline power before and after the load transfer is used as the amplitude of the sudden power change. This avoids calculation errors caused by instantaneous spike noise and dynamic power drift, quantifies the load fluctuation intensity of a single perturbation, and provides an accurate amplitude basis for calculating node correlation. Load transfer events correspond to a positive step rise, and load shedding events correspond to a negative step fall. The sign of the amplitude also contains event type information.

[0038] In identifying the affiliation between electricity meters and smart terminals in meter boxes, the line load switching events reported by the electricity meters are compared with those reported by the smart terminals in each meter box. If two events simultaneously meet three conditions—the same event type, an event occurrence time interval not exceeding a preset time interval threshold, and a relative deviation of the sudden power amplitude not exceeding a preset deviation threshold—they are considered to be matched. The number of matched events is counted. If a smart terminal in a meter box has a number of matched events that meet the preset matching quantity condition and are unique, then the electricity meter is determined to belong to that smart terminal. The formula for calculating the relative deviation of the sudden power amplitude is:

[0039] In the formula: , These are the two sudden power amplitudes corresponding to the same line load switching event reported by the meter and the smart terminal of the meter box.

[0040] This technical solution uses three constraints—event type, time interval, and relative amplitude deviation—to match the same source events between the meter and the smart terminal in the meter box. Consistent event types eliminate interference from dissimilar events; a time interval not exceeding a preset threshold (preferably 0.2 seconds) ensures consistency in the triggering time of the same disturbance event at upstream and downstream nodes through millisecond-level time synchronization; and a relative amplitude deviation not exceeding a preset threshold (preferably 5%) utilizes the physical law of load disturbances being transmitted in series along the line, attenuating at each level while maintaining similar amplitudes to eliminate cross-branch interference. Furthermore, by statistically analyzing the number of matched events and determining their uniqueness, the solution establishes the meter's affiliation to the meter box, avoiding incorrect matching issues across branch nodes.

[0041] The automatic topology construction and correction of the transformer substation in S6 also includes the identification of the affiliation between meter box smart terminals and distribution box smart terminals: assuming the total number of distribution box smart terminals in the network is... For smart terminals in meter boxes that report line load switching events no less than a preset active event threshold within a specified time period, the number of line load switching events that match with smart terminals in each distribution box is counted and recorded as follows: , For smart terminals in meter boxes and the first The number of line load switching events that match between the intelligent terminals of each distribution box within a specified time period; if there are intelligent terminals of each distribution box... If all three conditions are met, the smart terminal in the meter box is determined to belong to the smart terminal in the distribution box. : Condition 1, ,Right now The maximum value among the number of events matched between the smart terminal of the meter box and the smart terminals of each distribution box; Condition 2, ,in Set a preset threshold for the number of matches; Condition 3, It is unique, that is, except for the smart terminal of the distribution box. In addition, the number of matching events between the other distribution box smart terminals and the meter box smart terminals is less than [a certain number]. For nodes that remain on the pending list after identification, the master node selects a new time period and requires only the nodes on the pending list, as well as all smart terminals in meter boxes and distribution boxes across the network, to report line load switching events, and performs topology identification again. If there are still nodes that have not completed topology identification after the second identification, nodes with active electricity consumption are identified as nodes in different distribution areas; nodes with inactive electricity consumption are identified as nodes that cannot be identified based on line load switching events and are transferred to manual processing. This technical solution achieves the affiliation identification of the distribution box smart terminal by matching event count statistics and affiliation determination. It adopts the same triple condition matching method and unique affiliation determination logic as the meter-meter box level, realizing the bottom-up hierarchical splicing of the four-level topology. For silent nodes and weakly active nodes that fail to complete the same source matching in the first identification, the detection and matching are re-implemented by extending the detection time window and simplifying invalid monitoring data, thereby improving the topology identification coverage of the entire transformer area. For nodes that still cannot be identified after the second retest, nodes from different transformer areas and nodes that require manual processing are distinguished according to the degree of electricity consumption activity. This avoids the pollution of the topology results by interfering nodes from different transformer areas and ensures the complete closed loop of the identification process.

[0042] Compared with existing technologies, the distinguishing technical features of this invention are as follows: ① It uses the existing HPLC communication network to complete the millisecond-level synchronous acquisition of voltage and current at four monitoring nodes: distribution transformer, distribution box, meter box, and electricity meter. It calculates active power using integer-cycle discrete integration and extracts transient change features using DB5 orthogonal wavelet four-layer tower decomposition, providing event parameters with high signal-to-noise ratio and transient scale for matching of homogeneous events; ② Based on millisecond-level time synchronization, it performs cross-level homogeneous event matching with triple constraints of "event type-time interval-amplitude relative deviation" at the meter-meter box and meter box-distribution box levels, and introduces a dual-layer threshold joint judgment mechanism of three-level high-frequency details and four-level high-frequency details to suppress single-scale noise false triggering; ③ It completes the membership judgment of meter box to distribution box and electricity meter to meter box level by level with three conditions: maximum number of matched events, not less than the preset matching number threshold, and uniqueness. It performs secondary retesting and correction on nodes that are not matched for the first time by extending the detection window, forming a four-level topology that is pieced together from bottom to top and forms a complete closed loop. The overall technical solution based on the aforementioned distinguishing features is not a simple superposition of existing natural disturbance matching methods. Instead, it takes "synchronous acquisition - transient feature extraction - cross-layer homogeneous matching - uniqueness judgment - secondary retesting and correction" as the main line, and solves the technical problems that single-level natural disturbance matching cannot distinguish the four-level topology hierarchy and cannot completely restore the topology in the active area of ​​strong disturbance. It has outstanding substantive features and significant progress.

[0043] The present invention also provides a low-voltage distribution transformer area topology identification system for load switching event detection.

[0044] A low-voltage distribution transformer area topology identification system for load switching event detection includes: The data acquisition module performs high-frequency synchronous acquisition of voltage and current at each level of monitoring nodes, including distribution transformers, distribution boxes, meter boxes, and meters, within the distribution area through the HPLC communication network, and completes time synchronization of all nodes through the network time synchronization mechanism of the HPLC communication network. The signaling and time window management module generates detection signaling and sends it to all sub-nodes in the network when a topology identification requirement occurs in the transformer area, and configures a time window for a specified time period for the sub-nodes; The load switching event detection module, after receiving the signaling from the master node, calculates the active power time series based on the synchronously collected voltage and current time series data within a specified time period, uses discrete wavelet transform to extract transient change features from the active power time series, uses steady-state baseline difference method to calculate the change power amplitude, detects line load switching events, and reports event parameters. The topology identification module performs source matching based on the line load switching event parameters reported by all network nodes, completes the identification of the affiliation relationship between nodes at each level of meters, meter boxes, and distribution boxes, and includes nodes that have not completed source matching in the pending list. It completes secondary retesting and identification by extending the detection window to correct the topology identification results.

[0045] This technical solution, through the collaborative work of the data acquisition module, signaling and time window management module, load switching event detection module, and topology identification module, completely maps the automatic topology identification process of the transformer area into a modular system architecture: the data acquisition module and the signaling and time window management module together constitute the organization layer of the topology identification task, the load switching event detection module constitutes the distributed event perception layer, and the topology identification module constitutes the centralized topology inference layer. The modules are organically linked through the HPLC communication network. The system can be integrated into the transformer area fusion terminal or distributed and deployed at monitoring nodes at all levels, realizing full-process automation from electrical quantity acquisition, event detection, source matching to topology output, supporting the normalized online identification and dynamic updating of transformer area topology.

[0046] The one or more technical solutions provided by this invention have the following advantages compared with the prior art: (1) The equipment modification cost is extremely low and the project is highly feasible: the existing HPLC smart meters, branch monitoring terminals, integrated terminals and existing communication and measurement systems deployed in the distribution area are fully reused. There is no need to add special hardware such as signal injection modules and pulse generators. There is no need to replace equipment or modify lines. It is suitable for the modification and upgrading of various existing low-voltage distribution areas.

[0047] (2) No power quality interference from the power grid and high operational safety: The natural load switching disturbance generated by the daily power equipment of the users in the distribution area is used as the topology identification excitation source. There is no artificial harmonic or pulse signal output. It will not generate signal superposition and harmonic interference with power electronic equipment such as photovoltaic inverters, charging piles, and frequency conversion loads, thus ensuring the stability of power quality in the distribution area.

[0048] (3) It can accurately distinguish the topology level and support dynamic updates: By utilizing the physical law of load disturbance being transmitted in series along the line and attenuating step by step, it can not only distinguish the subordinate relationship of the same branch, but also distinguish the parent and child node levels of the upstream and downstream, and completely restore the four-level three-dimensional topology of "transformer-distribution box-meter box-meter". Through secondary retesting and rolling verification, it can automatically sense the changes in the transformer area structure and realize the dynamic and refined management of topology data. Attached Figure Description

[0049] Figure 1 This is a flowchart illustrating the overall process principle of the present invention.

[0050] Figure 2 This is a schematic diagram of the installation of electricity meters in a multi-meter box.

[0051] Figure 3 This is a schematic diagram of the four-level topology structure of the present invention: "transformer-distribution box-meter box-meter".

[0052] Figure 4 This is a schematic diagram of the DB5 wavelet scaling function and wavelet function.

[0053] Figure 5 This is a schematic diagram of the four-layer wavelet tower decomposition of the active power time series.

[0054] In the diagram: 1. Distribution transformer; 2. Distribution box; 3. Meter box; 4. Electricity meter; 5. Main node; 6. Sub-node. Detailed Implementation

[0055] Example 1 Please refer to Figure 1 and Figure 2 This embodiment provides a method for topology identification of a low-voltage distribution transformer area for load switching event detection, applied to a residential distribution transformer area. The area includes one distribution transformer 1, four distribution boxes 2, 27 meter boxes 3, and 305 user meters 4. A master node 5 is set on the low-voltage side of the distribution transformer 1, and sub-nodes 6 are set at each distribution box 2, meter box 3, and meter 4, forming a four-level hierarchical structure from top to bottom: "transformer-distribution box-meter box-meter". The master node 5 and each sub-node 6 are interconnected through the HPLC communication network deployed in the area.

[0056] like Figure 2 As shown, in the four-level topology structure of this embodiment, the low-voltage side outgoing lines of the distribution transformer 1 are sequentially connected to each distribution box 2. The outgoing lines of each distribution box 2 are connected to multiple meter boxes 3. Multiple user meters 4 are installed inside each meter box 3. All the lines of the distribution boxes 2 are connected to the low-voltage A, B, and C three-phase outgoing lines of the distribution transformer 1, forming a tree-like four-level topology architecture in series. The master node 5 is set on the low-voltage side of the distribution transformer 1 and is responsible for sending detection signals, summarizing the event parameters of the entire network, and performing topology identification. The child nodes 6 are set at each distribution box 2, meter box 3, and meter 4 and are responsible for collecting the voltage and current timing data of their respective locations and reporting the event parameters.

[0057] S1. Preparation for detecting load switching events in transformer substations: Through the HPLC communication network already deployed in the low-voltage distribution transformer area, high-frequency synchronous acquisition of voltage and current is performed on the monitoring nodes at each level, including distribution transformer 1, distribution box 2, meter box 3, and meter 4. Time synchronization of all nodes in the network is achieved through the network time synchronization mechanism of the HPLC communication network. In this embodiment, each monitoring node at each level performs high-frequency synchronous acquisition of the line voltage and current timing data at its location. The synchronization accuracy between nodes reaches the millisecond level, providing a precise time reference for matching cross-level load switching events. When a topology identification requirement arises in the transformer area, the master node 5 sends a signal to all sub-nodes 6 in the network through the HPLC communication network, requesting sub-nodes 6 to perform line load switching event detection within a specified time period and report the detection results to the master node 5. In this embodiment, the specified time period is one day.

[0058] S2. Construction of active power time series: Sub-node 6 calculates active power based on synchronously acquired discrete voltage and current sampling sequences, using a sliding window approach and an integer-cycle discrete integration algorithm. The active power calculation formula is as follows:

[0059] In the formula: This represents the number of discrete sampling points within a single power frequency cycle. , These are the first and second units within the power frequency cycle. The instantaneous voltage and current values ​​at each sampling time are used, with the number of sliding points being the preset window sliding step size. The calculation is advanced frame by frame through the sliding window to generate a continuous and smooth active power time series, preserving the transient fluctuation characteristics at the moment of load switching.

[0060] S3. Load switching transient event detection: like Figure 4 As shown, this embodiment uses the DB5 orthogonal wavelet to extract transient change features from the active power time series. The DB5 wavelet is one of the Daubechies series of orthogonal wavelets, and its scaling function and wavelet function have finite support and orthogonality properties. Figure 5 As shown, a four-level pyramid decomposition is performed on the active power time series: the first level decomposes the original power series into a first-level low-frequency approximation component and a first-level high-frequency detail component; the second level iterates the first-level low-frequency component again to obtain a second-level low-frequency approximation component and a second-level high-frequency detail component, and so on up to the fourth level. The decomposition iteration calculation formula is as follows:

[0061]

[0062] In the formula: This is the original active power time series. For the first Low-frequency approximation components of the layer, , The first Low-frequency approximate components and high-frequency detail components of the layer. These are the low-pass filter coefficients corresponding to the scaling function. These are the high-pass filter coefficients corresponding to the wavelet function. The number of decomposition layers, This represents the filter coefficient sequence number. After decomposition, four layers of high-frequency detail components are obtained, each with distinct physical scale differences: the shallow high-frequency components correspond to high-frequency measurement noise and minor load random fluctuations, while the deep high-frequency components correspond to mid-to-low frequency, longer-duration, and more stable equipment step switching disturbances.

[0063] Thresholds are set for the Level 3 and Level 4 high-frequency detail components. If the Level 3 and Level 4 high-frequency detail components simultaneously exhibit significant peak values ​​exceeding their respective thresholds at the same time point, then that time point is determined to be the actual load disturbance time point. Event classification is completed by combining the original power time series change trend: if the power time series corresponding to the disturbance time point shows a stable step upward trend, it is determined to be a load input event; if the power time series corresponding to the disturbance time point shows a stable step downward trend, it is determined to be a load shedding event.

[0064] S4. Standardized calculation of event parameters: After determining the effective load switching event, the amplitude of the sudden change power is calculated using the baseline difference method before and after the steady state of the event: taking the disturbance time T0 of wavelet positioning as the reference, the data of the preset steady state window is extracted forward, and the data of the steady state window after the disturbance is completed is extracted backward. The steady state baseline power before the switch and the steady state baseline power after the switch are calculated respectively. The formula for calculating the amplitude of the sudden change power is as follows:

[0065]

[0066]

[0067] In the formula: The steady-state baseline power before switching, This represents the steady-state baseline power after switching. For precise wavelet positioning of event moments. , These represent the lengths of the initial and final steady-state windows, respectively; the final mutation power amplitude is... For load input events, meet the requirements. Power increases positively by a step; for load shedding events, it satisfies... The power drops negatively by a step.

[0068] After detecting a line load switching event, sub-node 6 reports the parameters corresponding to this event to master node 5. The parameters include the event occurrence time, event type (load switching or load transfer), and sudden power amplitude. Table 1 shows the line load switching event parameters detected by a sub-node within a specified time period in this embodiment.

[0069] Table 1 Examples of Line Load Switching Event Parameters

[0070] S5, Cross-level same-origin event matching: After the specified time period ends, master node 5 performs topology identification based on the line load switching events reported by all nodes in the network. The identification of the affiliation between the electricity meter and the smart terminal in the meter box is as follows: Assume there are K smart terminals in the entire network. For electricity meter Y, let S be the number of line load switching events it reports within the specified time period. If S≤1, the meter is determined to be an inactive user and added to the pending list. If S≥2, the line load switching events reported by meter Y are compared with the load switching events reported by each of the K smart terminals in the network. For an event A from meter Y and an event B from the smart terminal in the meter box, if the three conditions are met simultaneously—the event type is the same, the time interval between events is no greater than 0.2 seconds, and the relative deviation of the sudden power amplitude of the two events is no greater than 5%—then the two events are considered to match. The formula for calculating the relative deviation is:

[0071] In the formula: , These are the two sudden power amplitudes corresponding to the same line load switching event reported by the meter and the smart terminal of the meter box.

[0072] Calculate the number of load switching events that match between meter Y and the K smart terminals in the meter box, denoted as N_{Y,k}. If a smart terminal k can be found that simultaneously meets the three conditions of maximum matching number, matching number not less than the preset matching number threshold, and uniqueness, then meter Y is determined to belong to smart terminal k. If no such terminal is found, meter Y is added to the pending list.

[0073] like Figure 5 As shown, the load switching event detection and reporting process is as follows: Sub-node 6 collects voltage and current data at high frequency, calculates the active power time series through integer-cycle discrete integration, detects load switching events through DB5 wavelet four-layer tower decomposition and dual-layer threshold joint judgment, calculates the sudden power amplitude through steady-state baseline differential, and finally reports the event occurrence time, event type, and sudden power amplitude to master node 5.

[0074] S6. Automatic construction and correction of transformer area topology: Identifying the affiliation between meter box smart terminals and distribution box smart terminals: Assume there are a total of [number missing] distribution box smart terminals in the entire network. One, smart terminal for meter box In this context, let the number of line load switching events reported within a specified time period be... ,like If so, the meter box is determined to be an inactive meter box and added to the pending list; if Then, the same method as for electricity meters is used to count the smart terminals in the meter boxes. The number of line load switching events that are matched with the intelligent terminals of each distribution box is denoted as , For smart terminals in meter boxes With the The number of line load switching events that match between the intelligent terminals of each distribution box within a specified time period; if there are intelligent terminals of each distribution box... All three conditions must be met simultaneously: Condition 1, ,Right now For smart terminals in meter boxes The maximum value among the number of events matched with the smart terminals of each distribution box; Condition 2, ,in Set a preset threshold for the number of matches; Condition 3, It is unique, that is, except for the smart terminal of the distribution box. In addition, the remaining distribution box smart terminals and meter box smart terminals The number of matching events is less than Then determine the smart terminal of the meter box. Belongs to the intelligent terminal of the distribution box If not found, then the smart terminal in the meter box will be... Added to the pending list.

[0075] After completing the identification of the two-level affiliation between the meter and meter box, and the meter box and distribution box, if the list to be processed is empty, it means that all nodes in the network have achieved topology identification. If the list to be processed is not empty, the master node 5 selects another time period (e.g., 2 days) to perform a supplementary topology identification for the nodes in the list to be processed. During this supplementary identification process, the master node 5 only requires the nodes in the list to be processed, as well as the smart terminals of the meter boxes and distribution boxes in the entire network, to report line load switching events, and then performs the same topology relationship judgment based on the number of matching events. If there are still nodes that have not completed topology identification after the two identifications, nodes with active electricity consumption behavior are determined to be nodes in different distribution areas, and nodes with inactive electricity consumption behavior are determined to be unable to achieve topology identification based on line load switching events, and are transferred to manual processing.

[0076] The automatic identification process for the transformer substation topology is as follows: Master node 5 summarizes the event parameters of the entire network, first performs the identification of the affiliation between the meter and the smart terminal of the meter box, then performs the identification of the affiliation between the smart terminal of the meter box and the smart terminal of the distribution box, extends the detection window of the nodes in the list to be processed for a second retest, and outputs the four-level topology structure of the transformer substation.

[0077] The usage process of the above embodiment is as follows: When a topology identification requirement arises in the transformer area, the master node 5 sends a signal to the sub-nodes 6 of the entire network through the HPLC communication network. The sub-nodes 6 collect voltage and current data at high frequency within a specified time period (1 day), calculate the active power time series through integer-cycle discrete integration, determine the load switching events of the detected lines through DB5 wavelet four-layer tower decomposition and dual-layer thresholding, calculate the amplitude of sudden power changes through steady-state baseline differential calculation, and report the event occurrence time, event type, and amplitude of sudden power changes. After the specified time period ends, the master node 5 summarizes the event parameters of the entire network, sequentially performs meter-meter box and meter box-distribution box two-level membership identification, extends the detection window of the nodes in the pending list for secondary retesting, and finally outputs the complete four-level tree topology structure of the transformer area. The entire process uses the natural load switching disturbances generated by the start and stop of users' daily electrical equipment as the excitation source, without the need for manual signal injection, additional hardware modification, or power outage work, and can realize 7×24-hour online autonomous identification and dynamic updating of the transformer area topology.

[0078] Example 2 Please refer to Figure 1 and Figure 5 This embodiment provides a low-voltage distribution transformer area topology identification system for load switching event detection, which is applied to the low-voltage distribution transformer area topology identification method for load switching event detection described in Embodiment 1. It includes the following modules: a data acquisition module, a signaling and time window management module, a load switching event detection module, and a topology identification module.

[0079] The data acquisition module performs high-frequency synchronous acquisition of voltage and current data from various monitoring nodes within the distribution area, including transformer 1, distribution box 2, meter box 3, and meter 4, via the HPLC communication network. It also achieves time synchronization across all nodes through the network time synchronization mechanism of the HPLC communication network. This module reuses the existing acquisition functions of the deployed HPLC smart meters, branch monitoring terminals, and integrated terminals in the distribution area, eliminating the need for additional hardware acquisition equipment. Each node independently acquires the voltage and current timing data of its location, achieving millisecond-level synchronization accuracy.

[0080] When a topology identification requirement arises in a transformer area, the signaling and time window management module generates detection signaling and distributes it to all sub-nodes in the network. It configures a time window for a specified period for each sub-node. This specified time window is a pre-set, limited-duration detection window, typically one day, which can be extended to two days for secondary testing. This module, through its on-demand triggering and window-based execution detection organization method, ensures that sufficient load switching event samples are collected within the time window while avoiding data redundancy and communication overhead caused by long-term continuous collection.

[0081] After receiving the signaling from the master node, the load switching event detection module calculates the active power time series based on the synchronously acquired voltage and current time series data within a specified time period. The active power calculation formula is as follows:

[0082] In the formula: This represents the number of discrete sampling points within a single power frequency cycle. , These are the first and second units within the power frequency cycle. The instantaneous voltage and instantaneous current values ​​at each sampling time are used, and the number of sliding points is the preset window sliding step size.

[0083]

[0084]

[0085] In the formula: This is the original active power time series. For the first Low-frequency approximation components of the layer, , The first Low-frequency approximate components and high-frequency detail components of the layer. These are the low-pass filter coefficients corresponding to the scaling function. These are the high-pass filter coefficients corresponding to the wavelet function. The number of decomposition layers, This refers to the filter coefficient index.

[0086]

[0087]

[0088]

[0089] In the formula: The steady-state baseline power before switching, This represents the steady-state baseline power after switching. For precise wavelet positioning of event moments. , These represent the lengths of the initial and final steady-state windows, respectively; the final mutation power amplitude is... For load input events, meet the requirements. Power increases positively by a step; for load shedding events, it satisfies... The power drops negatively by a step.

[0090] The topology identification module performs source matching on line load switching events reported by nodes at each level based on the line load switching event parameters reported by nodes across the entire network. A relative deviation formula is used when matching the electricity meter with the smart terminal in the meter box.

[0091] In the formula: , These are the two sudden power amplitudes corresponding to the same line load switching event reported by the meter and the smart terminal of the meter box.

[0092] and

[0093] In the formula: For smart terminals in meter boxes and the first The number of matching events for each smart terminal of the distribution box. For smart terminals in the meter box and the whole network The number of matching events corresponding to each smart terminal of the distribution box. This is a preset threshold for the number of matches.

[0094] In actual deployment, the data acquisition module is deployed at each level of monitoring node, the load switching event detection module is deployed at each sub-node or nearby edge computing device, and the signaling and time window management module and topology identification module are deployed at the integrated terminal of the distribution area where the master node is located. All modules are interconnected via the HPLC communication network and can also be integrated into the integrated terminal for unified operation. The system can continuously collect distribution area load switching event data over a long period, continuously verify and iteratively optimize the topology structure, automatically detect changes in distribution area structure such as user capacity expansion and cancellation, line upgrades, and new photovoltaic installations, and correct the topology ledger in real time, achieving dynamic and refined management of distribution area topology data.

[0095] Example 3 To verify the accuracy of the transformer substation topology identification method of this invention, with the support of Shandong Power Grid Company, a residential transformer substation in Qingdao City was selected for performance testing. This substation has 4 distribution boxes 2, 27 meter boxes 3, and 305 user meters 4. The method described in Example 1 was used to perform line load switching event detection and topology identification within a specified time period of one day.

[0096] During the test, all network nodes completed millisecond-level time synchronization through the HPLC communication network. Each sub-node calculated the active power time series based on the synchronously acquired data. After detecting line load switching events through DB5 wavelet four-layer tower decomposition and dual-layer threshold joint judgment, the master node performed cross-level homogeneous matching based on three conditions: event type, time interval (≤0.2 seconds), and relative amplitude deviation (≤5%), and completed the identification of the affiliation relationship between meter boxes and distribution boxes, and between meters and meter boxes layer by layer.

[0097] The test results are shown in Table 2: The topology identification results between the meter box and the distribution box were all accurate, with an accuracy rate of 100%. Of the 305 user meters, the topology identification results between 303 meters and the meter box were correct. The remaining 2 meters were for users with zero electricity consumption; no line load switching events were detected within the specified time period, making topology identification impossible based on load switching events. Therefore, topology identification was ultimately completed manually. The overall topology identification accuracy rate (excluding manually assisted nodes) reached 99.3%.

[0098] Table 2. Topology identification results of a residential radio station area in Qingdao.

[0099] To verify the advantages of this invention over existing technologies, based on the Qingdao transformer substation described in Example 3, this invention is compared horizontally with three typical existing methods: (a) Identification method based on steady-state time-series data correlation (corresponding to CN116436152A): relies on steady-state characteristic correlation, which is easily destroyed in high photovoltaic penetration transformer substations. It can only cluster within the same phase and substation, and cannot distinguish between upstream and downstream levels. The complete identification rate of the four-level topology is significantly reduced; (b) Active carrier injection identification method (corresponding to CN117277567A): requires the addition of a signal injection module and generates power quality disturbances, and does not utilize natural load disturbances; (c) Household transformer relationship identification method based on single-level natural appliance start-stop (corresponding to CN112234605A): can only achieve sub-meter-total single-level household transformer relationship matching, and cannot construct a four-level cross-layer topology. The comparison shows that, under the premise of reusing existing HPLC, eliminating the need for injection hardware, and avoiding power quality interference, this invention achieves complete and accurate automatic identification of the four-level topology (overall identification accuracy rate of 99.3%) by relying on the transient mutation features extracted by integer-cycle discrete integration and DB5 wavelet four-level decomposition, as well as cross-level homology matching, uniqueness judgment, and secondary retesting closed loop. In contrast, methods (a) and (b) are limited by new energy disturbances or hardware modifications, and method (c) is limited by single-level matching and cannot completely restore the four-level three-dimensional topology. This shows that this invention has a non-obvious overall architecture and a significant improvement in identification effect compared to the closest prior art CN112234605A.

[0100] The verification results of this embodiment show that the present invention can achieve high-precision automatic identification of transformer area topology through the natural load switching disturbances generated by the start and stop of users' daily electrical equipment. It does not require manual signal injection, additional hardware modification, or power outage operations, and has strong anti-disturbance capability for active transformer areas with high penetration of new energy and drastic load fluctuations. For extreme nodes such as zero power consumption and weak activity, the identification process is fully closed-loop through secondary retesting and manual backup mechanisms. The identification results can be directly used as basic data support for core distribution network businesses such as lean line loss management, accurate fault location, three-phase load imbalance management, and distributed photovoltaic consumption optimization.

[0101] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solutions of the present invention, and are not intended to limit the specific implementation of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the claims of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. A method for identifying the topology of a low-voltage distribution transformer area based on load switching event detection, characterized in that, Includes the following steps: S1. Preparation for load switching event detection in the distribution area: Through the HPLC communication network already deployed in the low-voltage distribution area, high-frequency synchronous acquisition of voltage and current is carried out on the monitoring nodes at all levels of distribution transformers, distribution boxes, meter boxes, and meters in the distribution area, and the time synchronization of all network nodes is completed through the network time synchronization mechanism of the HPLC communication network. S2. Active power time series construction: Within a specified time period, the sub-nodes calculate the active power using a sliding window method and an integer-cycle discrete integration algorithm based on the synchronously acquired voltage and current time series data, and generate an active power time series. S3. Load switching transient event detection: Discrete wavelet transform is used to extract transient change features from the active power time series to detect line load switching events; S4. Event Parameter Standardization Calculation: Based on the disturbance time located by the transient mutation feature extraction, the steady-state baseline difference method is used to calculate the mutation power amplitude, forming event parameters that include the event occurrence time, event type, and mutation power amplitude, and then reporting them to the master node. S5. Cross-level same-source event matching: The master node matches the line load switching events reported by nodes at each level based on the line load switching event parameters reported by all sub-nodes in the network. S6. Automatic Construction and Correction of Transformer Area Topology: Based on the matching results, the subordinate relationships between nodes at each level of meters, meter boxes, and distribution boxes are identified to obtain the transformer area topology structure. Nodes that have not been matched are included in the list to be processed. Secondary retesting and identification are completed by extending the detection window to correct the topology identification results.

2. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 1, characterized in that: In the preparation for the detection of load switching events in the distribution area in S1, the distribution area monitoring nodes include a main node set on the low-voltage side of the distribution transformer, and sub-nodes set at the distribution box, meter box, and electricity meter, forming a four-level hierarchical structure from top to bottom: "transformer-distribution box-meter box-electricity meter"; each level of monitoring node performs high-frequency synchronous acquisition of the line voltage and current timing operation data at its location, and the synchronization accuracy between nodes reaches the millisecond level.

3. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 1, characterized in that: In the preparation for the load switching event detection of the transformer area in S1, when a topology identification requirement occurs in the transformer area, the master node sends a signal to all sub-nodes in the network through the HPLC communication network, requiring the sub-nodes to perform line load switching event detection within a specified time period and report the detection results to the master node. The specified time period is a pre-set limited duration detection window.

4. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 1, characterized in that: In the construction of the active power time series of S2, the sub-nodes calculate the active power based on the synchronously acquired discrete voltage and current sampling sequences, using a sliding window method and an integer-cycle discrete integration algorithm. The active power calculation formula is as follows: In the formula: This represents the number of discrete sampling points within a single power frequency cycle. , These are the first and second units within the power frequency cycle. The instantaneous voltage and instantaneous current values ​​at each sampling time are used, and the number of sliding points is the preset window sliding step size.

5. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 1, characterized in that: In the load switching transient event detection of S3, discrete wavelet transform is used to extract transient change features from the active power time series; DB5 orthogonal wavelet is used to perform a four-level tower decomposition of the active power time series, extracting low-frequency approximate components and high-frequency detail components layer by layer: the first level decomposition yields first-level low-frequency approximate components and first-level high-frequency detail components; the second level further decomposes the first-level low-frequency approximate components to yield second-level low-frequency approximate components and second-level high-frequency detail components; the third level further decomposes the second-level low-frequency approximate components to yield third-level low-frequency approximate components and third-level high-frequency detail components; the fourth level further decomposes the third-level low-frequency approximate components to yield fourth-level low-frequency approximate components and fourth-level high-frequency detail components; the decomposition iteration calculation formula is: In the formula: This is the original active power time series. For the first Low-frequency approximation components of the layer, , The first Low-frequency approximate components and high-frequency detail components of the layer. These are the low-pass filter coefficients corresponding to the scaling function. These are the high-pass filter coefficients corresponding to the wavelet function. The number of decomposition layers, This refers to the filter coefficient index.

6. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 5, characterized in that: In the load switching transient event detection of S3, thresholds are set for the third-level high-frequency detail component and the fourth-level high-frequency detail component. If the third-level high-frequency detail component and the fourth-level high-frequency detail component simultaneously show significant peak values ​​exceeding their respective thresholds at the same time, then the time is determined to be the actual load disturbance time. The event is classified by combining the original power time series change trend: if the power time series corresponding to the disturbance time shows a stable step upward trend, it is determined to be a load switching event; if the power time series corresponding to the disturbance time shows a stable step downward trend, it is determined to be a load shedding event.

7. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 1, characterized in that: In the standardized calculation of event parameters in S4, the baseline difference method before and after the event steady state is used to calculate the amplitude of the sudden change power: taking the disturbance time located by the extraction of transient change features as the benchmark, a preset steady-state window of data is extracted forward, and a steady-state window of data after the disturbance is completed is extracted backward. The steady-state baseline power before the switch and the steady-state baseline power after the switch are calculated respectively. The formula for calculating the amplitude of the sudden change power is: In the formula: The steady-state baseline power before switching, This represents the steady-state baseline power after switching. For precise wavelet positioning of event moments. , These represent the lengths of the initial and final steady-state windows, respectively; the final mutation power amplitude is... For load input events, meet the requirements. Power increases positively by a step; for load shedding events, it satisfies... The power drops negatively by a step.

8. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 7, characterized in that: The cross-level same-source event matching in S5 includes the identification of the affiliation relationship between the meter and the smart terminal in the meter box: The line load switching events reported by the meter are compared with the line load switching events reported by each smart terminal in the meter box. If a line load switching event of the meter and a line load switching event of the smart terminal in the meter box simultaneously meet three conditions—the same event type, an event occurrence time interval not exceeding a preset time interval threshold, and a relative deviation of the sudden power amplitude of the two events not exceeding a preset deviation threshold—then the two events are determined to be matched. The number of line load switching events that are matched between the meter and each smart terminal in the meter box is counted. If there exists a smart terminal in the meter box whose number of matched events with the meter meets the preset matching quantity condition and is unique, then the meter is determined to belong to that smart terminal in the meter box. The formula for calculating the relative deviation of the two sudden power amplitudes is: In the formula: , These are the two sudden power amplitudes corresponding to the same line load switching event reported by the meter and the smart terminal of the meter box.

9. The low-voltage distribution transformer area topology identification method for load switching event detection as described in claim 1, characterized in that: The automatic topology construction and correction of the transformer substation in S6 also includes the identification of the affiliation between meter box smart terminals and distribution box smart terminals: assuming the total number of distribution box smart terminals in the network is... For smart terminals in meter boxes that report line load switching events no less than a preset active event threshold within a specified time period, the number of line load switching events that match with smart terminals in each distribution box is counted and recorded as follows: , For smart terminals in meter boxes and the first The number of line load switching events that match between the intelligent terminals of each distribution box within a specified time period; if there are intelligent terminals of each distribution box... If all three conditions are met, the smart terminal in the meter box is determined to belong to the smart terminal in the distribution box. : Condition 1, ,Right now The maximum value among the number of events matched between the smart terminal of the meter box and the smart terminals of each distribution box; Condition 2, ,in Set a preset threshold for the number of matches; Condition 3, It is unique, that is, except for the smart terminal of the distribution box. In addition, the number of matching events between the other distribution box smart terminals and the meter box smart terminals is less than [a certain number]. ; For nodes that remain in the pending list after identification, the master node selects a new time period and requires only the nodes in the pending list, as well as the smart terminals in the meter boxes and distribution boxes of the entire network, to report line load switching events, and then performs topology identification again. If any nodes still fail to complete topology identification after re-identification, nodes with active power consumption are identified as nodes in different power distribution zones; nodes with inactive power consumption are identified as nodes that cannot be identified based on line load switching events and are then handled manually.

10. A low-voltage distribution transformer area topology identification system for load switching event detection, employing the low-voltage distribution transformer area topology identification method for load switching event detection as described in any one of claims 1-9, characterized in that, Includes the following modules: The data acquisition module performs high-frequency synchronous acquisition of voltage and current at each level of monitoring nodes, including distribution transformers, distribution boxes, meter boxes, and meters, within the distribution area through the HPLC communication network, and completes time synchronization of all nodes through the network time synchronization mechanism of the HPLC communication network. The signaling and time window management module generates detection signaling and sends it to all sub-nodes in the network when a topology identification requirement occurs in the transformer area. It also configures a time window for a specified time period for the sub-nodes. The specified time period is a pre-set limited-duration detection window. The load switching event detection module, after receiving the signaling from the master node, calculates the active power time series based on the synchronously collected voltage and current time series data within a specified time period, uses discrete wavelet transform to extract transient change features from the active power time series, uses steady-state baseline difference method to calculate the change power amplitude, detects line load switching events, and reports event parameters. The topology identification module performs source matching based on the line load switching event parameters reported by all network nodes, completes the identification of the affiliation relationship between nodes at each level of meters, meter boxes, and distribution boxes, and includes nodes that have not completed source matching in the pending list. It completes secondary retesting and identification by extending the detection window to correct the topology identification results.

Images