A power distribution network data quality management and control method based on edge computing

Abstract

The present application relates to the technical field of edge computing, and particularly to a power distribution network data quality management and control method based on edge computing, wherein an edge node first acquires local data, generates an electrical consistency factor by checking the switch-current, upstream and downstream voltage and the conservation relationship of the total distribution table, determines a neighborhood mutual verification time window according to a time credibility factor, compares the same type of measurement data collected by the neighborhood edge nodes in the same area through the edge side nearest communication, and obtains a neighborhood mutual verification factor. Subsequently, the edge node fuses the electrical, neighborhood, time and link factors to calculate the comprehensive quality credibility; based on the credibility, the data is divided into repairable, isolated or pending review states, and local repair, local blocking or uploading to the main station for review are respectively performed. The method effectively improves the real-time quality of data.

Description

Technical Field

[0001] This invention relates to the field of edge computing technology, specifically to a method for managing and controlling data quality in power distribution networks based on edge computing. Background Technology

[0002] As the automation level of distribution networks continues to improve, a large number of smart meters, smart switch terminals, feeder terminals and monitoring equipment are connected to the power Internet of Things. The data sources for distribution network operation analysis, line loss management, lean operation and maintenance of transformer areas and load forecasting are increasingly dependent on on-site measurement data and equipment status data.

[0003] However, due to factors such as the quality of field communication links, abnormal timing systems, faulty acquisition terminals, and unsynchronized topology changes, the measurement data of the distribution network generally suffers from problems such as timestamp drift, missing data, incorrect connections, and inconsistencies between the status and actual operation. This makes it difficult to guarantee the reliability of the data in the main station side business system, resulting in phenomena such as distorted calculation results and false alarms or missed alarms.

[0004] In existing technologies, data quality verification is usually completed centrally on the main station side, through statistical analysis of historical data or consistency verification of the entire network topology. However, this has shortcomings such as reliance on centralized computing resources, high requirements for communication bandwidth, inability to utilize the proximity correlation between edge nodes for mutual verification, and limited support capabilities for real-time services.

[0005] Meanwhile, with the application and promotion of edge computing technology in power distribution networks, edge nodes have certain local computing and local communication capabilities, but there is still a lack of a data quality control mechanism that combines power distribution network electrical topology constraints, time synchronization status, neighborhood mutual verification, and communication link records to form a closed loop on the edge side, and to perform hierarchical access and local repair of data in order to improve data quality.

[0006] To address this, a data quality control method for power distribution networks based on edge computing is proposed. Summary of the Invention

[0007] The purpose of this invention is to provide a data quality control method for distribution networks based on edge computing. Edge nodes first acquire local data and generate an electrical consistency factor by verifying the balance between switch-current, upstream and downstream voltage, and the total meter readings of the distribution area. Based on a time reliability factor, a neighborhood verification time window is determined. The edge node compares similar measurement data collected by neighboring edge nodes in the same distribution area through nearby communication to obtain a neighborhood verification factor. Subsequently, the edge node integrates four factors—electrical, neighborhood, time, and link—to calculate a comprehensive quality reliability. Based on this reliability, the data is categorized into repairable, isolated, or pending verification states, and local repair, local blocking, or upload to the master station for verification are performed accordingly. This method effectively improves the real-time quality of the data.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A data quality control method for distribution networks based on edge computing includes: edge nodes acquiring locally collected measurement data to be verified, switch status data, communication link status data, and time synchronization status data; performing consistency verification on the measurement data to be verified based on pre-configured electrical consistency constraint rules to obtain an electrical consistency factor; the electrical consistency constraint rules include switch-current on / off relationship, upstream and downstream voltage gradient relationship, and the conservation relationship between the main and sub-meters in the distribution area; Edge nodes determine the time reliability factor based on the time synchronization status, and determine the neighborhood mutual verification time window based on the time reliability factor; within the neighborhood mutual verification time window, they acquire similar measurement data collected by neighboring edge nodes in the same area through nearby communication on the edge side, and compare the measurement data to be verified with the similar measurement data based on the neighborhood mutual verification weight to obtain the neighborhood mutual verification factor; the edge nodes determine the link reliability factor based on the communication link status of the communication link corresponding to the measurement data to be verified. Edge nodes integrate four factors: electrical consistency factor, neighborhood mutual verification factor, time reliability factor, and link reliability factor, to obtain a comprehensive quality reliability score. Based on the comprehensive quality reliability score, the business access status is determined, which includes repairable status, isolated status, and pending review status. For data in a repairable state, perform local repair; for data in an isolated state, perform local blocking; for data in a pending review state, upload the original measurement data of the data to be reviewed, as well as the corresponding comprehensive quality reliability and the above four factors, to the main site.

[0009] Preferably, the measurement data to be verified includes voltage measurement data and meter data; the meter data includes the total meter data and sub-meter data for the distribution area; the configuration process of the electrical consistency constraint rules includes: the edge node obtains the topology version identifier and topology connection relationship of the distribution network, determines the feeder position, distribution area position, switch position and meter connection position corresponding to the edge node, and generates the corresponding electrical consistency constraint rules for the topology position based on the feeder position, distribution area position, switch position and meter connection position.

[0010] Preferably, the generation process of the electrical consistency factor includes: edge nodes determining whether the current on / off state of the circuit associated with the switch matches the switch status data to obtain the switch-current on / off verification result; determining whether the voltage change direction conforms to the topology power supply direction based on the voltage measurement data of upstream and downstream nodes to obtain the upstream and downstream voltage gradient verification result; determining whether the power conservation relationship meets the preset conservation conditions based on the total meter data and sub-meter data of the transformer area to obtain the transformer area total and sub-meter conservation verification result; and edge nodes integrating the switch-current on / off verification result, the upstream and downstream voltage gradient verification result, and the transformer area total and sub-meter conservation verification result to obtain the electrical consistency factor.

[0011] Preferably, the generation process of the time reliability factor includes: the edge node acquiring the local clock state, the time source state, the most recent time transmission record and the data acquisition timestamp, determining the time reliability factor based on the degree of deviation between the local clock and the time source, and adjusting the neighborhood mutual verification time window based on the time reliability factor.

[0012] Preferably, the process of determining the neighborhood mutual verification time window includes: when the time credibility factor indicates that the time synchronization state is credible, the edge node uses a first time window for neighborhood mutual verification; when the time credibility factor indicates that the time synchronization state is abnormal, the edge node uses a second time window for neighborhood mutual verification; wherein, the time range covered by the second time window is greater than the time range covered by the first time window.

[0013] Preferably, the generation process of the neighborhood mutual verification factor includes: edge nodes determining the measurement type and topology location corresponding to the measurement data to be verified; screening neighboring edge nodes that have an electrical relationship with the topology location within the same transformer area; acquiring the same type of measurement data collected by the neighboring edge nodes within the neighborhood mutual verification time window through edge-side proximity communication; comparing the deviation and trend of the measurement data to be verified with the same type of measurement data; and generating the neighborhood mutual verification factor based on the deviation comparison results and trend comparison results.

[0014] Preferably, the process of determining the neighborhood mutual verification weight includes: the edge node calculates the initial weight corresponding to each neighborhood edge node through a preset weight calculation function based on the electrical distance, communication delay, and data acquisition time difference between the neighborhood edge node and the corresponding measurement point of the measurement data to be verified, and normalizes each initial weight to obtain the neighborhood mutual verification weight; wherein, the initial weight is negatively correlated with the electrical distance, the communication delay, and the data acquisition time difference.

[0015] Preferably, the process of generating the link trust factor includes: the edge node reading the communication link record through which the measurement data to be verified is transmitted from the acquisition terminal to the edge node, the communication link record including link connectivity status, transmission delay, retransmission record, packet loss record and link switching record, and the edge node determining the link trust factor based on the communication link record.

[0016] Preferably, the process of determining the service access status includes: edge nodes fusing electrical consistency factors, neighborhood mutual verification factors, time reliability factors, and link reliability factors according to preset fusion rules to obtain a comprehensive quality reliability; when the comprehensive quality reliability meets the local repair access conditions, the service access status is determined to be a repairable status; when the comprehensive quality reliability meets the service blocking conditions, the service access status is determined to be an isolated status; when the comprehensive quality reliability does not meet the local repair access conditions and does not meet the service blocking conditions, the service access status is determined to be a pending review status.

[0017] Preferably, the process of local repair for data in a repairable state includes: the edge node selecting a corresponding repair method based on the anomaly type of the measurement data to be verified, the repair method including timestamp correction, data compensation for adjacent acquisition cycles, compensation for similar measurement data in the neighboring area, and data correction based on the electrical consistency constraint rules; the edge node recalculates the comprehensive quality reliability of the repaired data, and provides the repaired data to the target service when the recalculated comprehensive quality reliability meets the target service access conditions.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. In existing technologies, abnormal time synchronization mechanisms can cause data acquisition timescales at edge nodes to shift, thereby affecting the time matching relationship between local measurement data and neighboring measurement data. This invention converts the time synchronization state into a time reliability factor and dynamically selects a neighboring mutual verification time window based on this factor. When the time synchronization state is reliable, a first time window is used, and neighboring measurement data of the same type with a small timescale difference from the local measurement data to be verified are selected for mutual verification. When the time synchronization state is abnormal, a second time window with a larger coverage area is used to expand the search range of candidate neighboring data.

[0019] Under the second time window, this invention increases the weight of trend comparison results in the similarity score and decreases the weight of single-point deviation comparison results; for neighboring nodes with insufficient time matching or for which effective time-series data cannot be obtained, they are not included in the calculation of the neighbor mutual verification factor. This reduces the impact of natural load changes over a longer time span on the neighbor mutual verification results and improves the accuracy of the neighbor mutual verification results under abnormal time synchronization conditions.

[0020] 2. This invention retains the calculation results of the electrical consistency factor, neighborhood mutual verification factor, time reliability factor, and link reliability factor, and combines them with data missing measurement markers, switch-current on / off verification results, upstream and downstream voltage gradient verification results, and transformer area total and sub-meter conservation verification results to determine the anomaly type corresponding to the measurement data to be verified. For anomalies that can be uniquely attributed, edge nodes use adjacent acquisition cycle data compensation, timestamp correction, neighborhood similar measurement data compensation, data correction based on electrical consistency constraint rules, or complete retransmission of successful data and complete data after link switching to correct the transmitted data. For cases where the cause of the anomaly cannot be uniquely determined or where effective repair data is lacking, the corresponding data is transferred to a pending review state. This improves the interpretability and determinism of anomaly cause identification, local repair, and service access processes.

[0021] 3. Traditional data verification heavily relies on the main station to centrally aggregate data from the entire network for verification, which not only places extremely high demands on communication bandwidth but also struggles to support real-time service requirements. This invention decentralizes pre-defined "topology electrical consistency rules" to the local edge nodes and pulls neighboring data from the same area. For data without missing measurement markers and where all factors can be calculated normally, this invention obtains a comprehensive quality reliability by fusing EF, NF, TF, and LF at the edge, and determines the service access status based on the specific values ​​of each factor. For data with missing measurements, verification results pending confirmation, or relevant factors that cannot be reliably calculated, data is diverted according to data compensation or pending verification conditions, thereby determining the measurement data as repairable, isolated, or pending verification at the edge. This solution not only directly blocks inferior data and repairs salvageable data at the edge, effectively intercepting invalid traffic uploads, but also allows edge nodes to directly output high-quality data to downstream services such as "line loss analysis" and "lean operation and maintenance." Attached Figure Description

[0022] Figure 1 A flowchart illustrating a data quality control method for power distribution networks based on edge computing, provided in an embodiment of the present invention; Figure 2 A schematic diagram illustrating the configuration process of an electrical consistency constraint rule provided in an embodiment of the present invention; Figure 3 A schematic diagram illustrating the generation process of an electrical consistency factor according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the generation of a time reliability factor and the determination of a neighborhood mutual verification time window, as provided in an embodiment of the present invention. Figure 5 A schematic diagram illustrating the generation process of neighborhood mutual verification factor and neighborhood mutual verification weight provided in an embodiment of the present invention; Figure 6A schematic diagram illustrating the generation process of a link trust factor according to an embodiment of the present invention; Figure 7 This is a schematic diagram of a comprehensive quality reliability fusion, business access, and local repair process provided in an embodiment of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figures 1 to 7 This invention provides a data quality control method for distribution networks based on edge computing. This method is executed by edge nodes deployed on the distribution network substation side. The edge nodes establish communication connections with data acquisition terminals, smart switch terminals, neighboring edge nodes within the same substation area, and the master station. A detailed description follows.

[0025] A data quality control method for distribution networks based on edge computing includes: edge nodes acquiring locally collected measurement data to be verified, switch status data, communication link status data, and time synchronization status data; performing consistency verification on the measurement data to be verified based on pre-configured electrical consistency constraint rules to obtain an electrical consistency factor; the electrical consistency constraint rules include switch-current on / off relationship, upstream and downstream voltage gradient relationship, and the conservation relationship between the main and sub-meters in the distribution area; Edge nodes determine the time reliability factor based on the time synchronization status, and determine the neighborhood mutual verification time window based on the time reliability factor; within the neighborhood mutual verification time window, they acquire similar measurement data collected by neighboring edge nodes in the same area through nearby communication on the edge side, and compare the measurement data to be verified with the similar measurement data based on the neighborhood mutual verification weight to obtain the neighborhood mutual verification factor; the edge nodes determine the link reliability factor based on the communication link status of the communication link corresponding to the measurement data to be verified. Edge nodes integrate four factors: electrical consistency factor, neighborhood mutual verification factor, time reliability factor, and link reliability factor, to obtain a comprehensive quality reliability score. Based on the comprehensive quality reliability score, the business access status is determined, which includes repairable status, isolated status, and pending review status. For data in a repairable state, perform local repair; for data in an isolated state, perform local blocking; for data in a pending review state, upload the original measurement data of the data to be reviewed, as well as the corresponding comprehensive quality reliability and the above four factors, to the main site.

[0026] Combination Figure 1This is a flowchart illustrating a data quality control method for power distribution networks based on edge computing, provided by an embodiment of the present invention. The specific implementation process of this embodiment is as follows (steps S101 to S110): S101: Data Acquisition: Edge nodes acquire local measurement data (such as voltage, current, active power, reactive power, etc.) at a fixed period (15 minutes per acquisition period in this embodiment; for switch and status quantities, an event-triggered + second-level heartbeat method can be used). Simultaneously, they acquire switch status data ("on / off" status and its flip-over time stamp) via the communication interface with the switch intelligent terminal. They also synchronously record the connectivity status, latency, and other raw parameters of the current acquisition link; and record the local clock status and time source status. These four types of data are packaged with a unified acquisition time stamp to form a single data record to be processed.

[0027] S102: Electrical Consistency Verification (Generate EF): The edge node calls the pre-configured electrical consistency constraint rules (including three types of sub-rules: switch-current on / off relationship, upstream and downstream voltage gradient relationship, and transformer area total meter conservation relationship) to verify the measurement data to be verified collected in S101, and obtains the electrical consistency factor EF, with the value range normalized to [0,1].

[0028] S103: Time Trust Factor and Neighborhood Mutual Verification Time Window (Generate TF): The edge node calculates the time trust factor TF based on the local clock state, time source state, the most recent time synchronization record, and the data acquisition timestamp of S101; and selects the neighborhood mutual verification time window from the preset two-level time window according to the TF.

[0029] S104: Neighborhood Mutual Verification (Generate NF): Within the neighborhood mutual verification time window determined in S103, the edge node requests the same type of measurement data from neighboring edge nodes with electrical relationships within the same area through the nearest edge communication. Combining the neighborhood mutual verification weight, the local data and the neighboring data are compared for deviation and trend to obtain the neighborhood mutual verification factor NF.

[0030] S105: Link Trustworthiness Assessment (LF Generation): The edge node reads the link records (connectivity status, transmission delay, retransmission records, packet loss records, link switching records) that the measurement data to be verified passes through from the acquisition terminal to the edge node, and obtains the link trustworthiness factor LF.

[0031] S106: Four-Factor Fusion (Generation T). For data without a missing data marker, and where the electrical consistency factor EF, neighborhood verification factor NF, time reliability factor TF, and link reliability factor LF can all be calculated normally, the edge node fuses the above four factors according to the preset fusion rules to obtain the comprehensive quality reliability T∈[0,1]. For data with a missing data marker, verification results pending confirmation, or any factor that cannot be reliably calculated, the formal four-factor fusion in this step is not performed, and the business access status is determined according to the rules described in S107 and S702.

[0032] S107: Business Access Status Determination. For data for which no valid measurements were collected in this data collection cycle, edge nodes are marked as missing data, and the overall quality reliability is not calculated. Based on whether there is valid measurement data from adjacent acquisition cycles or valid neighboring measurement data of the same type that can be used for compensation, the business access status of the data is determined to be either repairable or pending review.

[0033] For valid measurements already collected in this acquisition cycle, and the electrical consistency factor Neighborhood mutual verification factor Time reliability factor and link trust factor For data that can be calculated normally, the edge node determines the business access status of the data as repairable, isolated, or pending review status based on the comprehensive quality reliability T and the specific value combination of each component factor.

[0034] For the switch-current continuity test results Upstream and downstream voltage gradient verification results Marked as pending confirmation, or , , , If any factor in the data cannot be reliably calculated, the data will not be automatically determined to be in an isolated or repairable state based on the temporary calculation results, but will instead be determined to be in a state pending review.

[0035] S108~S110: Branch processing. Figure 1 S106 to S110 in the document outline the general steps for comprehensive quality reliability fusion, business access status determination, and branch processing. Figure 7 S701 to S709 in the above steps are detailed processes.

[0036] in, Figure 1 S106 in the middle corresponds to Figure 7 S701 in the text is used to integrate the electrical consistency factor. Neighborhood mutual verification factor Time reliability factor and link trust factor To obtain the overall quality reliability ; Figure 1 S107 in the middle corresponds to Figure 7 S702 in the document is used to determine the business access status; Figure 1 S108 in Figure 7 S703 to S707 are used to determine the anomaly type, select the repair method, perform local repair and post-repair review, and provide the repaired data to the target business when it meets the target business access conditions. Figure 1 S109 in Figure 7 The S708 in the code is used to perform local blocking on data in an isolated state; Figure 1 S110 in the middle corresponds to Figure 7 S709 in the system is used to upload data to be reviewed or data that still does not meet the access conditions of the target business after repair to the main site.

[0037] For data in a repairable state, edge nodes follow... Figure 7 S703 to S707 perform anomaly type determination, repair method selection, local repair, comprehensive quality reliability calculation after repair, target business access condition review, and repair data output. When the repaired data meets the target business access conditions, the repaired data is provided to the target business through S707. When the repaired data does not meet the target business access conditions, the corresponding data is uploaded to the main station for review through S709.

[0038] For isolated data, edge nodes follow... Figure 7 The S708 in the system performs a local blocking action, marking the data as unavailable and stopping the provision of the data to local downstream services.

[0039] For data in the pending review status, edge nodes are configured according to... Figure 7 S709 uploads the original measurement data, the obtained comprehensive quality confidence level T, and the values ​​of EF, NF, TF, and LF factors to the main station, and uploads data with missing measurements, verification pending confirmation, factors that cannot be calculated, or repair failures. For comprehensive quality confidence levels and related factors that have not been calculated or cannot be reliably calculated, no mandatory values ​​are assigned.

[0040] Furthermore, the measurement data to be verified includes voltage measurement data and meter data; the meter data includes the total meter data and sub-meter data for the distribution area; the configuration process of the electrical consistency constraint rules includes: the edge node obtains the topology version identifier and topology connection relationship of the distribution network, determines the feeder position, distribution area position, switch position and meter connection position corresponding to the edge node, and generates the corresponding electrical consistency constraint rules for the topology position based on the feeder position, distribution area position, switch position and meter connection position.

[0041] Specifically, such as Figure 2 A schematic diagram illustrating the configuration process of an electrical consistency constraint rule provided in an embodiment of the present invention; S201: After initialization or topology change, edge nodes obtain the currently effective topology version identifier and topology connection relationship data through the interface with the master station's topology service. The topology connection relationship data is given in the form of an adjacency list, with each record including at least the node number, node type, parent node number, list of child node numbers, rated voltage level, equivalent impedance parameters of adjacent line segments, and rated capacity parameters of the corresponding branch. The equivalent impedance parameters include the equivalent resistance and equivalent reactance of the line segment, and the rated capacity parameters are the rated apparent power capacity of the corresponding line or distribution equipment. These parameters are distributed to the edge nodes by the master station's topology service based on the distribution network design data, equipment ledger, and the currently effective topology version, and are then stored and updated locally by the edge nodes according to the topology version identifier.

[0042] S202: The edge node performs a matching search in the topology connection relationship data based on its own device number to determine the feeder location (feeder number), transformer area location (transformer area number), switch location (the number of the switch device it directly monitors or is associated with, if any), and meter connection location (the connection node of the meter it directly collects in the topology tree, i.e., whether its superior is the main meter node or a branch meter node).

[0043] S203: Based on the topology location information determined in S202, the edge node generates a set of electrical consistency constraint rules specific to this node according to a preset rule template, including: Switch-current on / off relationship rules: If S202 determines that the switching equipment associated with this node is connected, then open status verification rules and closed status verification rules are generated respectively.

[0044] When the switch is in the open position, the measured current of the associated circuit should not exceed the leakage current threshold. When the measured current is consistently greater than the leakage current threshold At that time, it is determined that there is a strong physical conflict between the switch state and the current on / off state of the associated circuit.

[0045] When the switch is in the closed state, the measured current of the associated line should, in principle, not be lower than the starting threshold. When the measured current is lower than the start-up threshold When the switch status does not directly match the current on / off status of the associated line, the condition is marked as a closed low load condition, and further judgment is made by combining the switch status and measured current values ​​in the most recent three consecutive valid acquisition cycles.

[0046] If the switch status remains closed for the most recent three consecutive valid acquisition cycles and there is no switch status reversal, the closed low load state will be treated as a normal no-load or light-load condition. If the switch status reversal occurs, the number of valid acquisition cycles is insufficient, or it cannot be determined whether the low current is caused by normal no-load or light-load, the switch-current on / off verification result will be marked as pending confirmation.

[0047] and The current is set proportionally according to the rated current of the branch where the switch is located. In this embodiment, In this embodiment, 2% of the rated current is used. In this embodiment, the rated current is taken as 0.5%.

[0048] Upstream and downstream voltage gradient relationship rules: Edge nodes determine the direct upstream node number, rated voltage, and allowable voltage drop limit of their own nodes based on the topological connection relationship, and generate upstream and downstream voltage gradient relationship rules.

[0049] In the absence of distributed power sources supplying power in reverse to the current branch, the voltage difference between the upstream node and the current node should be within a preset allowable voltage drop range. In this embodiment, the upper limit of the allowable voltage drop is taken as 7% of the rated voltage of the line segment.

[0050] In cases where there is reverse power supply from distributed sources in the current branch or the current power flow direction cannot be determined, instead of directly using the fixed upstream-to-downstream voltage gradient relationship to determine the measurement data as abnormal, the upstream and downstream voltage gradient verification results are marked as pending confirmation, and the corresponding data is transferred to the pending review state.

[0051] The rule for maintaining the balance between main and sub-meters in a transformer area: If S202 determines that the meter connection location of this node is a main meter or a sub-meter, then extract the list of all sub-meter node numbers under the same transformer area from the topology connection data; the comprehensive line loss rate of a transformer area is statistically calculated by the main station's line loss management system according to the transformer area and the settlement cycle, and uses a single, determined value corresponding to the current settlement cycle. The data is distributed to the local storage of the edge nodes. The 3% to 5% range refers to the historical variation range of the line loss rate for different settlement cycles in this embodiment, and does not indicate that the edge nodes can choose the line loss rate within this range.

[0052] When performing the conservation check of the distribution area total table, the edge node directly reads the specific line loss rate value corresponding to the current distribution area identifier and the current settlement cycle. and in accordance with Calculate and estimate line loss electricity. If the specific value of the line loss rate corresponding to the current settlement period is not obtained, the line loss rate of the most recent valid settlement period is used, and the calculation result is marked with "line loss rate continued"; if there is no valid line loss rate, the protection check of the total meter and sub-meter of the transformer area for this period is not performed, and the protection check result of the total meter and sub-meter of the transformer area is marked as uncalculated.

[0053] When an effective line loss rate exists, the edge node reads the line loss rate value corresponding to the current transformer area from the line loss rate data stored locally, and generates the rule that "the total meter power consumption should be approximately equal to the sum of the power consumption of all sub-meters plus the line loss power consumption estimated according to the line loss rate, and the deviation rate between the two should be less than the preset threshold".

[0054] S204: Edge nodes will store the generated electrical consistency constraint rules locally in the form of a rule table, which can be directly called by the S102 electrical consistency verification step; when the topology version identifier changes, S201 to S204 will be executed repeatedly to complete the dynamic update of the rules.

[0055] like Figure 3 A schematic diagram illustrating the generation process of an electrical consistency factor according to an embodiment of the present invention; Furthermore, the generation process of the electrical consistency factor includes: edge nodes determining whether the current on / off state of the circuit associated with the switch matches the switch status data to obtain the switch-current on / off verification result; determining whether the voltage change direction conforms to the topology power supply direction based on the voltage measurement data of upstream and downstream nodes to obtain the upstream and downstream voltage gradient verification result; determining whether the power conservation relationship meets the preset conservation conditions based on the total meter data and sub-meter data of the transformer area to obtain the transformer area total and sub-meter conservation verification result; and edge nodes integrating the switch-current on / off verification result, the upstream and downstream voltage gradient verification result, and the transformer area total and sub-meter conservation verification result to obtain the electrical consistency factor.

[0056] S300: The edge node reads the data required for this electrical consistency verification from the measurement data to be verified and the switch status data obtained from S101, and combines it with the topology constraint rules that have been stored locally in S204 as the input for verification of S301 to S303.

[0057] S301: Switch-current continuity verification. The edge node reads the switch status data and the measured current value of the corresponding branch within the current acquisition period. And make a judgment according to the switch-current on / off relationship rules configured in S203.

[0058] When the switch is in the open state and ≤ When the switch state matches the current on / off state of the associated circuit, the switch-current on / off verification result is determined. =1.

[0059] When the switch is in the open state and > At this time, the edge node continues to read the switch status and measured current value for the most recent three consecutive valid acquisition cycles; if the switch status for the most recent three consecutive valid acquisition cycles is all in the open state, and the measured current value is greater than 100%, the edge node will continue to read the switch status and measured current value for the most recent three consecutive valid acquisition cycles. If there is a strong physical conflict between the switch state and the current on / off state of the associated circuit, the switch-current on / off verification result is determined. =0; If the above situation does not occur consecutively, then the switch-current on / off verification result will be 0. Marked as pending confirmation.

[0060] When the switch is in the closed state and ≥ When the switch state matches the current on / off state of the associated circuit, the switch-current on / off verification result is determined. =1.

[0061] When the switch is in the closed state but < At this time, the edge node marks this operating condition as a closed low-load state, and does not directly determine the switch-current on / off verification result as... =0.

[0062] For the aforementioned low-load closing condition, if the switch status remains closed for the most recent three consecutive valid acquisition cycles, and there is no switch status reversal, then it is treated as a normal no-load or light-load condition, and the switch-current continuity verification result is determined as follows. =1; If the number of valid acquisition cycles is insufficient, the switch state reverses, or it cannot be determined whether the low current is caused by normal no-load or light-load conditions, then the switch-current continuity test result will be reset. Mark it as pending confirmation and transfer the corresponding data to the pending review status described in S702.

[0063] S302: Upstream and downstream voltage gradient verification: Edge nodes acquire voltage measurements of their local nodes. Voltage measurements of its direct upstream nodes (determined by the topology connection described in S202) (Can be obtained from local historical cache or through neighborhood communication as described in S104), calculate voltage drop according to power supply direction. If 0≤ If the voltage drop is less than or equal to the upper limit of the allowable voltage drop (S203 configuration), then the direction and magnitude of the voltage change are determined to conform to the voltage gradient relationship when power is supplied from the upstream node to this node. The upstream and downstream voltage gradient verification results are then used. =1; if If the voltage drop exceeds the upper limit of the allowable voltage drop, the voltage change is determined to be inconsistent with the preset voltage gradient relationship. The upstream and downstream voltage gradient verification results are then used. =0.

[0064] like If the value is negative and exceeds the preset reasonable fluctuation range, the edge node further determines whether the current branch is experiencing a reverse power supply condition from the distributed power source. The reverse power supply condition is determined based on the transformer area operation status information issued by the master station, the positive and negative active power indicators, or the operation status of the distributed power source grid connection point.

[0065] When it is determined that there is no reverse power supply condition in the current branch, the upstream and downstream voltage gradient verification results are determined as follows. =0.

[0066] When it is determined that the current branch is experiencing reverse power transmission, or when the current power flow direction cannot be reliably determined, the current branch should not be directly connected to the power supply. A negative value is considered an anomaly in the measurement data, and the upstream and downstream voltage gradient verification results are then... Mark it as pending confirmation and transfer the corresponding data to the pending review status described in S702.

[0067] To avoid misjudgments caused by voltage measurement jumps within a single acquisition cycle, edge nodes perform analysis on the voltage measurements within the three most recent valid acquisition cycles. Take the median and use the median to perform the above judgment.

[0068] S303: Transformer Area Meter Consistency Verification. If the meter connection point of this node is a master meter or a sub-meter, the edge node, through the nearest edge-side communication method, requests or forwards electricity data level by level within the same transformer area according to the master meter-sub-meter connection hierarchy in the transformer area topology tree: For nodes connected to a master meter, it collects the electricity data reported by all sub-meter nodes within the transformer area; for nodes connected to a sub-meter, it reports the local electricity data to the master meter node or a pre-designated aggregation node within the transformer area. This data acquisition covers all metering points within the transformer area and is independent of and unrestricted by the range of neighboring nodes used for subsequent neighboring area mutual verification. Edge nodes (or master meter nodes undertaking aggregation functions) aggregate the transformer area master meter electricity within the same settlement cycle. Sum of all meter readings Line loss rate configured according to S203 Estimate line loss power Calculate the deviation rate .like If the value is less than a preset threshold (5% in this embodiment), it is determined that the conservation condition is met. =1; otherwise =0. When this node is neither a master meter nor a sub-meter node (e.g., only a voltage / switch monitoring point), The default value is 1 (this rule is not included in the verification of this node).

[0069] The settlement period in this embodiment is a calendar month; the edge node continuously accumulates the electricity increment of each metering point's sub-meter / main meter according to the 15-minute collection cycle described in S101, and maintains the rolling cumulative electricity value within this settlement period locally; when a settlement period is completed, or when a settlement trigger instruction is received from the main station, the edge node performs the following sub-meter conservation check based on the rolling cumulative value; the obtained value... The results remain valid within the current settlement period until the cumulative value is updated in the next settlement period.

[0070] S304: Weighted fusion yields Edge nodes are assigned preset weights. , , Perform weighted fusion: ,in .

[0071] Because the switch-current on / off relationship is a strong physical constraint, abnormalities often correspond to fundamental problems such as incorrect connections or hanging. Therefore, its weight can be set to a higher value, such as taking... =0.4, =0.3, =0.3; To determine the weights of each sub-rule used for weighted fusion, this invention adopts a step-by-step method based on historical data statistical analysis and operation and maintenance experience: First, select historical operating data for no less than one statistical period in the target distribution network, and collect measurement data, operation and maintenance records, and corresponding fault or false alarm tags; then, for each sub-rule, calculate the proportion of real anomalies detected and the false alarm rate when the sub-rule is triggered alone, and obtain the effectiveness index of the sub-rule; further, normalize each effectiveness index to obtain the initial weight, and the scores of operation and maintenance experts on the relative importance of each sub-rule can be multiplied by the above effectiveness index as coefficients, and then normalized again; finally, during the trial operation phase, the weights are iteratively adjusted according to the missed detection and false alarm situation. Through this method, suitable sub-rule weight configurations can be determined in different distribution networks, resulting in 0.4, =0.3, =0.3.

[0072] For sub-rules that do not participate in the verification (such as this node not being associated with switching equipment, or as in S303 where this node is neither a master table nor a sub-table node, etc.), (In the case where the default value is 1), the weights of the sub-rules participating in the actual verification at the edge nodes are re-normalized proportionally before weighted fusion: that is, when only , When participating in the verification, use , Replace each , The weighted fusion calculation in this step is performed; when only one sub-rule participates in the verification, the normalized weight of that sub-rule is set to 1. Through the above deterministic normalization rules, it is ensured that... The calculation method has a unique and definite calculation path for any node type, avoiding the impact of inherently missing constraints on performance. .

[0073] When S301 verifies the switch-current on / off status result When marked as pending confirmation, do not... The value is assigned to 0 or 1, and f1 is not used to participate in the formal weighted fusion of the electrical consistency factor EF.

[0074] Edge nodes retain the "Switch - Current On / Off Status Pending Confirmation" flag and renormalize the corresponding weights according to the other sub-rules that can be calculated normally, and calculate the temporary electrical consistency factor EF. The temporary electrical consistency factor EF is only used as auxiliary information for the master station's review and is not used to automatically determine the isolation status or repairability status of the data.

[0075] For data marked "Switch-current on / off status pending confirmation", the formal four-factor fusion described in S701 is not performed, but the pending verification status determination described in S702 is directly entered.

[0076] When S302 marks the upstream and downstream voltage gradient verification result f2 as pending confirmation, or S303 marks the transformer area total / splitter balance verification result f3 as uncalculable, the corresponding verification result is not assigned a value of 0 or 1, nor is the formal weighted fusion of the electrical consistency factor EF performed. The edge node retains the corresponding pending confirmation or uncalculable mark and transfers the data to the pending review state described in S702.

[0077] Figure 4 This is a schematic diagram illustrating the generation of a time reliability factor and the determination of a neighborhood mutual verification time window, as provided in an embodiment of the present invention. Furthermore, the generation process of the time reliability factor includes: the edge node acquiring the local clock state, the time source state, the most recent time transmission record and the data acquisition timestamp, determining the time reliability factor based on the degree of deviation between the local clock and the time source, and adjusting the neighborhood mutual verification time window based on the time reliability factor.

[0078] S401: Data Acquisition: Edge nodes continuously record: local clock status (whether the local crystal oscillator is running normally), time source status (whether BeiDou / GPS time synchronization is locked, or whether network time synchronization NTP / PTP is online), and the timestamp of the most recent successful time synchronization. and the local-to-reference deviation record of the time synchronization, and the current data acquisition timestamp. .

[0079] S402: Calculation of Time Reliability Factor Edge nodes calculate the cumulative deviation of the local clock relative to the timing reference. If the time source is in normal condition, the absolute value of the deviation between the local clock obtained in this time synchronization verification and the time reference is directly determined as the cumulative deviation. ; If the time synchronization source is in an abnormal state, the most recent time synchronization record will be used. Starting from the local crystal oscillator's nominal drift rate (2 ppm in this example, which can be provided by the equipment's factory parameters), the cumulative deviation is estimated. Drift rate.

[0080] Edge nodes determine their temporal reliability factors according to the following priorities and piecewise functions. Ensure that any (time source state) Each combination has a unique and definite value: (1) When the cumulative deviation If the time reliability factor is greater than 200ms, or if the time source remains in an abnormal state for a period exceeding the preset offline duration, the time reliability factor will be adjusted accordingly. =0.2. The preset offline duration is 10 minutes in this embodiment. This condition has a higher priority than the following conditions.

[0081] (2) If condition (1) is not met, when the time source is in normal condition and the cumulative deviation is... Time reliability factor when ≤50ms =1.0.

[0082] (3) If condition (1) is not met, when the timing source is in normal condition and 50ms < cumulative deviation When the time reliability factor (TF) is ≤200ms, it is calculated according to the following formula: =1.0 - [( [-50ms) / 150ms]×0.6.

[0083] According to the above formula, when the time synchronization source is in normal condition and = 50ms, TF = 1.0; when the time source is in normal status and =200ms, =0.4, TF varies with cumulative deviation in the range of 50ms to 200ms. It decreases linearly as it increases.

[0084] (4) If condition (1) is not met, when the timing source is in an abnormal state and the estimated cumulative deviation is... Time reliability factor when ≤50ms The value is fixed at 0.994.

[0085] This value is used to characterize that although the current estimated cumulative deviation does not exceed 50ms, the uncertainty of the estimation result is higher than that of the time synchronization source under normal conditions because the time synchronization source is already in an abnormal state. Therefore, when the time synchronization source is normal and ≤50ms Additional risk discounts are set on top of 1.0.

[0086] (5) If condition (1) is not met, when the timing source is in an abnormal state and 50ms < the estimated cumulative deviation When the time reliability factor (TF) is ≤200ms, it is calculated according to the following formula: TF = 0.994 - [( [-50ms) / 150ms]×0.594.

[0087] According to the above formula, when the time synchronization source is in an abnormal state and =50ms, =0.994; when the time synchronization source is in an abnormal state and =200ms, =0.4, TF varies with cumulative deviation in the range of 50ms to 200ms. It decreases linearly as it increases.

[0088] Using the above segmented calculation method, the value of the time reliability factor TF can be uniquely determined under various conditions, including when the time source is in normal condition, when the time source is in abnormal condition, when the cumulative deviation is in different intervals, and when the time source is continuously offline.

[0089] The process of determining the neighborhood mutual verification time window includes: when the time credibility factor indicates that the time synchronization state is credible, the edge node uses the first time window for neighborhood mutual verification; when the time credibility factor indicates that the time synchronization state is abnormal, the edge node uses the second time window for neighborhood mutual verification; wherein, the time range covered by the second time window is greater than the time range covered by the first time window.

[0090] S403~S405: Neighborhood Mutual Verification Time Window Selection: Edge nodes are preset with two levels of time windows, both of which are set based on the acquisition cycle (15 minutes) described in S101: The first time window (covering the most recent 3 acquisition cycles, i.e., ±45 minutes) is used to acquire similar measurement data of neighboring nodes within a few close acquisition cycles when the time synchronization status is reliable; the second time window (covering the most recent 5 acquisition cycles, i.e., ±75 minutes, with a significantly larger coverage range than the first time window) is used to compensate for the risk of neighboring data mismatch caused by the uncertainty of local time stamps when the time synchronization status is abnormal, with a larger time range. When the value is ≥0.7 (indicating the reliability of the time synchronization state), the first time window is used; when... When the value is <0.7 (indicating an abnormal time synchronization state), a second time window is used. To avoid... Slight fluctuations around the 0.7 threshold cause frequent time window switching. Edge nodes determine the time window level based on the TF values ​​of the three most recent consecutive acquisition cycles: only when... If the value is ≥0.7 for three consecutive acquisition cycles, switch to (or maintain) the first time window; only when If the value is less than 0.7 for three consecutive acquisition cycles, switch to (or maintain) the second time window; otherwise, maintain the current time window level.

[0091] To determine the threshold of the time reliability factor and the length of the two-level time windows, this embodiment employs a method based on historical data statistical analysis. Historical operational data within a certain time range is collected from the target distribution network, recording the time synchronization error, the value of the time reliability factor, and the corresponding time synchronization alarm status. For samples marked as "normal time synchronization," an empirical distribution of the time reliability factor is statistically analyzed, and a preset quantile (e.g., the 95th percentile) of this distribution is selected as the time reliability threshold. When the time reliability factor is greater than or equal to the time reliability threshold, the time synchronization state is considered reliable; when the time reliability factor is less than the time reliability threshold, the time synchronization state is considered abnormal. In this embodiment, the quantile was statistically obtained to be approximately 0.7 in a certain distribution network; therefore, the time reliability threshold is set to 0.7.

[0092] The lengths of the two-tiered time windows are determined based on the acquisition cycle and the duration of load disturbances. During the trial operation phase, edge nodes statistically analyze the duration distribution of typical load disturbances and timescale drift in historical measurement data. A time range covering most disturbance events is selected as the upper limit of the second time window length. The length of the first time window is then set considering computational complexity and real-time requirements. In this embodiment, a 15-minute acquisition cycle is used. Statistical results show that most disturbances end within 1-2 hours. Therefore, the first time window is set to cover the most recent 3 acquisition cycles (±45 minutes), and the second time window is set to cover the most recent 5 acquisition cycles (±75 minutes).

[0093] When the first time window is used, the edge node selects the neighboring measurement data of the same type with the smallest acquisition time difference with the local measurement data to be verified from the first time window, and uses it as the input data for subsequent deviation comparison.

[0094] When a second time window is used, it is used to expand the search range of candidate neighborhood data, but it does not mean that all neighborhood data within the second time window will directly participate in the deviation comparison. The edge node selects the neighborhood measurement data of the same type with the smallest acquisition time difference with the local measurement data to be verified from the second time window, and compares it with the temporal change trend of the neighborhood node within the second time window.

[0095] If a neighboring node does not have data that matches the local measurement data to be verified in time within the second time window, or cannot obtain time series data for no less than 3 valid acquisition cycles, the neighboring node is marked as having insufficient time matching, and the data of the neighboring node will not be used to calculate the current neighbor mutual verification factor NF.

[0096] Specifically, such as Figure 5 A schematic diagram illustrating the generation process of neighborhood mutual verification factor and neighborhood mutual verification weight provided in an embodiment of the present invention; Furthermore, the generation process of the neighborhood mutual verification factor includes: edge nodes determining the measurement type and topology location corresponding to the measurement data to be verified; screening neighboring edge nodes that have an electrical relationship with the topology location within the same transformer area; acquiring the same type of measurement data collected by the neighboring edge nodes within the neighborhood mutual verification time window through edge-side proximity communication; comparing the deviation and trend of the measurement data to be verified with the same type of measurement data; and generating the neighborhood mutual verification factor based on the deviation comparison results and trend comparison results.

[0097] The process of determining the neighborhood mutual verification weight includes: the edge node calculates the initial weight corresponding to each neighborhood edge node according to the electrical distance, communication delay and data acquisition time difference between the neighborhood edge node and the corresponding measurement point of the measurement data to be verified, through a preset weight calculation function, and normalizes each initial weight to obtain the neighborhood mutual verification weight; wherein, the initial weight is negatively correlated with the electrical distance, the communication delay and the data acquisition time difference.

[0098] Specifically, this includes: S501: Determine the measurement type and topology location: The edge node obtains the measurement type and topology coordinates (to which the transformer area, to which the branch, and topology level belong) required for this mutual verification based on the data type label (voltage / current / active power, etc.) of the measurement data to be verified and the topology location of the node determined in S202.

[0099] S502: Filtering Neighborhood Edge Nodes: In the same transformer area topology tree, filter out neighboring edge nodes that have an electrical relationship with the current node's topological location (i.e., adjacent connection points on the same branch, or collection points at both ends of the same line segment), forming a neighborhood candidate set. .

[0100] S503: Acquire similar data from the neighborhood. The edge node sends a data request to each neighboring node in the neighborhood candidate set through edge-side proximity communication, which includes HPLC or low-power wireless mesh. The data request is used to acquire similar measurement data collected by each neighboring node within the neighborhood mutual verification time window determined in S405.

[0101] Edge nodes set a preset communication timeout period from the moment a data request is sent. Neighboring nodes that return complete data within the preset communication timeout period, and whose returned data passes integrity verification and topology version verification, are identified as valid neighboring nodes.

[0102] For neighboring nodes that fail to return complete data within the preset communication timeout period, whose returned data fails the integrity check, or whose returned data corresponds to a topology version that is inconsistent with the currently effective topology version, they will be removed from the current valid neighboring candidate set, and the corresponding communication timeout or data verification failure flag will be recorded.

[0103] S504: Deviation Comparison and Trend Comparison: For the candidate set, the first... Neighboring nodes: During deviation comparison, the edge node converts the value collected by the i-th neighboring node into the expected value corresponding to the position of the current node based on the topological relationship.

[0104] For voltage measurement data, the edge node calculates the line voltage drop correction based on the equivalent impedance parameters of the line segment between the neighboring node and the current of the corresponding branch, and uses the line voltage drop correction to correct the voltage measurement value of the neighboring node. For current or energy data, the edge node normalizes the collected values ​​of the neighboring node based on the ratio of the rated capacity of the branch where the neighboring node is located to the rated capacity of the branch where the current node is located.

[0105] When the equivalent impedance parameter or branch rated capacity parameter of the corresponding line segment is missing, the version is inconsistent, or the preset validity period is exceeded, the data conversion of the neighboring node will not be performed, and the neighboring node will be removed from the current valid neighboring candidate set.

[0106] For valid neighboring nodes that have completed data conversion and have not been eliminated, the edge node selects the closest set of local measured values ​​and converted expected values ​​within the neighbor mutual verification time window, and calculates the deviation rate. |Local measured value - expected value| / rated value; Trend comparison: Take the time series of at least 3 collection cycles of the local and neighboring nodes within the mutual verification time window, and calculate the Pearson correlation coefficient. .

[0107] Edge nodes set the weights of deviation comparison results and trend comparison results in the similarity score according to the currently used neighborhood mutual verification time window.

[0108] When using the first time window, the calculation is performed according to the following formula. Similarity scores corresponding to each neighboring node ; ; When using the second time window, the calculation is performed according to the following formula. Similarity scores corresponding to each neighboring node : in, The preset upper limit for deviation rate normalization is 5% in this embodiment. The coverage of the second time window is larger than that of the first time window. Therefore, the weight of the trend comparison result is increased and the weight of the single-point deviation comparison result is decreased under the second time window to reduce the impact of natural load changes on the neighborhood mutual verification results over a longer time span.

[0109] when When <0.8 (considered as inconsistent trend), the first If a neighboring node is marked as having inconsistent trends, an "abnormal trend" label will be added to the pending review information uploaded to the main site via S709. If the neighboring node is also marked as having insufficient time matching, the similarity score corresponding to that neighboring node will not be calculated. And remove the neighboring node from the current set of valid neighboring candidate nodes.

[0110] S505: Neighborhood Mutual Verification Weight Calculation: The edge node determines the effective neighborhood candidate set V based on the data returned by S503 and the deviation comparison and trend comparison results of S504.

[0111] The neighboring nodes in the effective neighborhood candidate set V should simultaneously satisfy the following conditions: (1) Similar measurement data collected by the neighboring node within the neighboring mutual verification time window has been obtained; (2) The neighboring node was not marked as having insufficient time matching by S504; (3) The equivalent impedance parameters, branch rated capacity parameters and other data conversion parameters of the line segment corresponding to the neighboring node are complete and valid, and the parameter version is consistent with the currently effective topology version; (4) It can correctly calculate the deviation rate corresponding to the neighboring node. Pearson correlation coefficient and similarity score .

[0112] For the i-th neighbor node in the effective neighbor candidate set V, the edge node obtains its electrical distance. Neighborhood mutual verification communication delay (milliseconds) and data acquisition time difference (milliseconds), and calculate the initial weight corresponding to the neighboring node according to the following formula. ; in, This represents the electrical distance between the i-th neighboring node and the corresponding measurement point of the measurement data to be verified. In this embodiment, it is represented by the number of line segments that are separated in the topology tree. This represents the end-to-end communication delay between the i-th neighboring edge node and this edge node when performing neighboring mutual verification data requests and data returns. It is determined by this edge node based on the time difference between the time when the neighboring data request is sent and the time when the complete neighboring data is received. This represents the absolute value of the time difference between the data acquisition time of the i-th neighboring node and the local data acquisition time of the measurement to be verified. These are scaling coefficients used to normalize electrical distance, neighborhood mutual verification communication delay, and data acquisition time difference into dimensionless quantities, respectively. In this embodiment, they are taken as 1, 0.01, and 0.02, respectively.

[0113] To determine the aforementioned scaling factor, this invention statistically analyzed the numerical ranges of typical electrical distance, communication delay, and data acquisition time difference for each neighboring node over a period of time during the trial operation phase, and calculated their average or median values ​​respectively. and will Set it to the reciprocal of the above average, such that This ensures that each item has a similar magnitude of influence on the weights under typical operating conditions. In this embodiment, statistics from a certain distribution network are used... These correspond to 1, 100 milliseconds, and 50 milliseconds respectively, therefore we take... The values ​​are 1, 0.01, and 0.02.

[0114] The initial weight electrical distance Neighborhood mutual verification communication delay and data collection time difference All showed a negative correlation.

[0115] Edge nodes normalize the initial weights of each neighboring node in the effective neighboring candidate set V according to the following formula to obtain the neighboring mutual verification weight corresponding to the i-th neighboring node. , ; After normalization, the sum of the mutual verification weights of all neighboring nodes in the effective neighboring candidate set V is 1, that is... ; When some neighboring nodes fail to match due to insufficient time, missing data conversion parameters, inconsistent parameter versions, invalid data, or inability to calculate similarity scores... When a node is removed, the edge node removes the neighboring node from the effective neighboring candidate set V and recalculates the initial weight and normalized weight for the remaining effective neighboring nodes after the removal.

[0116] The removed neighboring nodes do not participate in the normalized weight calculation, nor do they participate in the weighted fusion of the neighbor mutual verification factor NF as described in S506.

[0117] When the effective neighborhood candidate set V is empty, the edge node does not perform initial weight calculation and normalization processing, marks the neighborhood mutual verification factor NF as uncomputable, and records the reason that the effective neighborhood candidate set is empty.

[0118] For data where the neighborhood mutual verification factor NF is not computable, NF is not simply assigned a value of 0, nor is this situation directly interpreted as a neighborhood mutual verification anomaly. Instead, the corresponding data is transferred to the pending verification state described in S702, and the main station performs verification by combining the neighborhood data request record, time matching result, and the obtained EF, TF, and LF factor values.

[0119] S506: Neighborhood mutual verification factor fusion.

[0120] When there is at least one valid neighbor node in the current valid neighbor candidate set, the edge node performs a weighted summation of the similarity scores corresponding to each valid neighbor node according to the normalized neighbor mutual verification weight, to obtain the neighbor mutual verification factor: .

[0121] When all neighboring nodes experience communication timeouts, return data verification failures, or topology version inconsistencies, resulting in an empty set of valid neighboring candidate nodes, the edge nodes do not perform neighbor mutual verification factor fusion, mark the neighbor mutual verification factor NF as uncomputable, and transfer the corresponding data to the pending verification state described in S702.

[0122] like Figure 6 A schematic diagram illustrating the generation process of a link trust factor according to an embodiment of the present invention; Furthermore, the process of generating the link trust factor includes: the edge node reading the communication link record that the measurement data to be verified passes through from the acquisition terminal to the edge node, the communication link record including link connectivity status, transmission delay, retransmission record, packet loss record and link switching record, and the edge node determining the link trust factor based on the communication link record.

[0123] Specifically, S601~S602: Read link records. The edge node reads the communication link record corresponding to this measurement data to be verified, including: link connectivity status, and acquisition link transmission delay. (milliseconds), number of retransmissions Packet loss records (packet loss rate during this collection period) ), Link switching records (number of link reselections / switches within this collection period) Among them, the transmission delay of the acquisition link. This represents the end-to-end transmission delay when the acquisition terminal transmits the measurement data to be verified to this edge node. It is determined by the edge node based on the time difference between the time the acquisition terminal sends the data and the time the complete measurement data is received.

[0124] S603: Calculate the penalty component: Calculate the normalized penalty component for each record item at the edge node (value range [0,1], 0 indicates no effect, 1 indicates severe effect): Chain break penalty If an interruption occurred during this cycle, set the value to 1; otherwise, set the value to 0. Delay penalty ,in The acceptable latency threshold is set at 100ms in this embodiment. The penalty saturation threshold is set (1000ms is selected in this embodiment); Retransmission penalty , Take 5 times; Packet loss penalty , Take 10%; Switching penalties , Take 3 times.

[0125] S604: Fusion obtained Edge nodes are fused according to preset weights for each penalty component: , .

[0126] Pick =0.35 (Broken links have the most direct impact on data credibility and have the highest weight). =0.15, =0.15, =0.20, =0.15. The calculation results are limited to the [0,1] interval.

[0127] The parameters involved in this embodiment, such as the voltage drop upper limit, leakage current protection threshold, activation threshold, link delay threshold, retransmission count threshold, and packet loss rate threshold, are all determined through statistical analysis of historical operating data and used as configurable parameters of the system. For the voltage drop upper limit, the voltage drop data of the user-end voltage relative to the upstream voltage of the transformer area under normal operating conditions is selected, its distribution is statistically analyzed, and the upper limit is selected in conjunction with national and industry standards. For the leakage current threshold and activation threshold, the leakage current level under normal operating conditions is statistically analyzed in conjunction with the setting principles of the grounding protection device, and a value slightly higher than the maximum normal leakage current is selected as the protection threshold, while currents significantly lower than this value are used as the activation criterion. For the link delay, retransmission count, and packet loss rate thresholds, the distribution of delay, retransmission count, and packet loss rate of the communication link under normal conditions is statistically analyzed, and the 95th or 99th percentile of the corresponding distribution is selected as the threshold. The above thresholds can be issued and adjusted by the master station or management system according to the transformer capacity, line characteristics and communication network type of different distribution areas. In this embodiment, the voltage drop limit is set to 7%, the leakage current protection threshold is set to 2% of the rated current, and the link delay threshold is set to 100 milliseconds and 1000 milliseconds, etc.

[0128] like Figure 7 This invention provides a schematic diagram of a comprehensive quality reliability fusion, business access, and local repair process. Furthermore, the process of determining the service access status includes: edge nodes fusing electrical consistency factors, neighborhood mutual verification factors, time reliability factors, and the link reliability factors according to preset fusion rules to obtain a comprehensive quality reliability; when the comprehensive quality reliability meets the local repair access conditions, the service access status is determined to be a repairable status; when the comprehensive quality reliability meets the service blocking conditions, the service access status is determined to be an isolated status; when the comprehensive quality reliability does not meet the local repair access conditions and does not meet the service blocking conditions, the service access status is determined to be a pending review status.

[0129] Furthermore, the process of local repair for data in a repairable state includes: the edge node selects the corresponding repair method according to the anomaly type of the measurement data to be verified, the repair method includes timestamp correction, adjacent acquisition cycle data compensation, neighboring similar type measurement data compensation, and data correction based on the electrical consistency constraint rules; the edge node recalculates the comprehensive quality reliability of the repaired data, and provides the repaired data to the target service when the recalculated comprehensive quality reliability meets the target service access conditions.

[0130] The specific process includes: S701: Four-factor fusion: Edge nodes are fused according to preset weights , , , : , .Pick , , , Electrical consistency factor and neighborhood mutual verification factor directly reflect the degree of consistency between the data itself and physical laws and neighborhood evidence, and are set with relatively high weights; time reliability factor and link reliability factor reflect the reliability of data acquisition and transmission process more, and are used as auxiliary correction terms with relatively low weights.

[0131] To determine the aforementioned weights, this embodiment employs a weight determination method based on historical data and operational experience: The values ​​of EF, NF, TF, and LF are recorded in the training dataset, along with whether the corresponding measurement data is ultimately determined to be valid or invalid, establishing a weight determination method in the form of... The linear model uses the above judgment results as labels and employs linear discriminant methods such as least squares fitting or logistic regression to solve for the weights. The obtained weights are normalized to ensure they are non-negative and sum to 1. During the trial operation phase, the weights are iteratively adjusted based on missed detections and false alarms. Through these steps, those skilled in the art can determine suitable weight configurations for different distribution networks. Therefore, in this embodiment, [the weight configuration is selected]. , , , .

[0132] When no valid measurement value is collected in the current acquisition cycle, the edge node sets a data missing mark and... and Data marked as incalculable. For data marked as missing, the overall quality confidence score is not calculated. And not implemented and , Instead of comparing, it directly enters the data missing test repair branch in S702; however, when the communication link record indicates that the link is continuously broken in this acquisition cycle and there is no historical data or neighboring data that can be used for compensation, the data is directly determined to be in a state pending review.

[0133] Edge nodes in computation At the same time, retain the sub-verification results of S304. , , as well as , , , The numerical values ​​of each component (including components marked as "uncalcifiable" and their marking states) are available for direct use in subsequent steps S702-S704, without needing to rely solely on scalar values. Inferring the causes from the opposite perspective.

[0134] S702: Business Access Condition Determination. Edge nodes determine the business access status of the measurement data to be verified in the following order: data missing status, business interruption conditions, local repair access conditions, and pending verification conditions.

[0135] (a) Determination of missing data status If no valid measurement value of the measurement data to be verified is collected in the current acquisition cycle, the edge node sets a data missing mark and records the electrical consistency factor that cannot be calculated due to the lack of valid measurement values. and neighborhood mutual verification factor Marked as not computable.

[0136] For data with missing data markers, edge nodes do not have their overall quality confidence score calculated. It also does not implement comprehensive quality reliability. With low confidence threshold Or high confidence threshold A comparison.

[0137] If there is valid measurement data from an adjacent acquisition period or valid neighboring measurement data of the same type that can be used to compensate for the missing measurement data, the service access status of the data is determined to be repairable; if there is no valid measurement data from an adjacent acquisition period or valid neighboring measurement data of the same type that can be used to compensate for the missing measurement data, the service access status of the data is determined to be pending review.

[0138] (ii) Business interruption conditions For data without a missing measurement marker, only when the switch is in the open state and the measured current value of the associated line is greater than the leakage current threshold in the most recent three consecutive valid acquisition cycles, will the missing measurement be considered. To verify the switch-current continuity test results Only when the value is 0 is a strong physical conflict between the switch state and the current on / off state of the associated circuit considered to exist; regardless of overall quality reliability. The value of , whether high or low, determines the business access status of the data as an isolated state.

[0139] When the switch is closed but the measured current of the associated line is lower than the starting threshold. The above does not apply to the low load condition of closing the circuit breaker. =0 Direct isolation condition. If determined by S301 to be a normal no-load or light-load condition, then... If it cannot be determined, then... Mark it as pending confirmation, and determine the business access status of the corresponding data as pending review.

[0140] exist If the overall quality reliability is 1, If so, the business access status of the data is determined to be isolated. Among these, the low confidence threshold... In this embodiment, 0.35 is used.

[0141] (III) Local Restoration Eligibility Criteria For data that has not been marked as missing and does not meet the conditions for service interruption, if it meets either the first or second condition below, the service access status of the data will be determined as repairable.

[0142] Category 1 condition: Overall quality reliability Among them, the high confidence threshold In this embodiment, 0.80 is used.

[0143] Second type of condition: Overall quality reliability satisfy And exactly one of the following four single attributable anomalies is met: 1. Time synchronization anomaly: Time reliability factor <0.7, and the switch-current continuity test result =1. Upstream and downstream voltage gradient verification results =1. Conservation verification result of the total score table of the transformer area =1, Neighborhood mutual verification factor NF≥0.6, Link trust factor ; 2. Neighborhood Mutual Validation Anomaly: Neighborhood Mutual Validation Factor <0.6, and the switch-current continuity test result =1. Upstream and downstream voltage gradient verification results =1. Conservation verification result of the total score table of the transformer area =1. Time reliability factor Link Trust Factor ; 3. The total score table for the transformer area shows an anomaly in conservation: The conservation check result for the total score table for the transformer area is as follows. =0, and the switch-current continuity test result =1. Upstream and downstream voltage gradient verification results =1, Neighborhood mutual verification factor NF≥0.6, Time reliability factor Link Trust Factor ; 4. Link transmission anomaly: Link reliability factor And the switch-current on / off verification results =1. Upstream and downstream voltage gradient verification results =1. Conservation verification result of the total score table of the transformer area =1, Neighborhood mutual verification factor NF≥0.6, Time reliability factor .

[0144] in, This indicates that time synchronization is normal, and NF ≥ 0.6 indicates that neighborhood mutual authentication is normal. This indicates that the communication link is in normal condition; when , or When the value of is equal to its corresponding threshold, it is processed as a normal state.

[0145] The phrase "just one of them" means that only one of the four anomalies mentioned above—time synchronization anomaly, neighborhood mutual verification anomaly, substation total table conservation anomaly, and link transmission anomaly—is true, while the other three anomalies are not true.

[0146] (iv) Conditions pending review For data without a missing data flag, if the data does not meet the business blocking conditions and the local repair access conditions, the business access status of the data will be set to pending review.

[0147] The pending review status includes at least the following situations: 1. The overall quality reliability T satisfies However, it may simultaneously trigger two or more abnormal situations. 2. The overall quality reliability T satisfies However, the cause of the anomaly cannot be uniquely determined; 3. Upstream and downstream voltage gradient verification results =0, but does not meet the service interruption conditions and there is no clear automatic repair path; or during the upstream and downstream voltage gradient verification process described in S302, it is determined that the current branch has a distributed power source reverse power supply condition, or the power flow direction of the current branch cannot be reliably determined, resulting in the upstream and downstream voltage gradient verification result being 0. Marked as pending confirmation; 4. During the switch-current continuity test described in S301, the switch is in a closed low-load state, or the switch is in an open state but the current exceeds... The situation did not occur consecutively, leading to errors in the switch-current continuity test results. Marked as pending confirmation; 5. Electrical consistency factor EF, neighborhood mutual verification factor The time reliability factor (TF) or link reliability factor (LF) cannot be calculated due to missing input parameters, inconsistent parameter versions, or invalid input data. 6. Data missing markers exist, but there are no valid measurement data from adjacent acquisition cycles or valid neighboring measurement data of the same type that can be used for compensation; 7. The link trust factor LF < 0.6, but complete retransmission success data, complete data after link switching, or effective neighborhood compensation data cannot be obtained; 8. Although the repair conditions are met, the necessary input data for the repair cannot be obtained according to S703 to S704, including situations where the target abnormal sub-table cannot be uniquely identified when the total sub-table conservation is abnormal, invalid line loss rate parameters, or inconsistent settlement cycles between the total table and the sub-table data.

[0148] S703~S704: Anomaly Type Determination and Repair Method Selection: Regardless of whether the repairable state is determined via " The path is still determined through a single attributable anomaly path. Edge nodes check the component results retained by S701 in the following order of priority to determine the specific anomaly type and select the corresponding repair method (if all components are normal, then the substantive repair is skipped and the path is directly released, which is equivalent to a no-operation): If a missing data marker is present, the data is considered missing.

[0149] When valid acquisition period data exist before and after the missing acquisition period, linear interpolation is performed using the data from the previous valid acquisition period and the data from the next valid acquisition period, and the interpolation result is used as the corrected measurement data.

[0150] When only valid acquisition period data exists before or after the missing acquisition period, the valid acquisition period data closest to the missing acquisition period is used for compensation.

[0151] When there is no valid adjacent acquisition period data, but there is a valid neighborhood expected value obtained according to S504, the valid neighborhood expected value is used for compensation.

[0152] When there is neither valid adjacent acquisition period data nor valid neighborhood conversion expectation value, automatic repair is not performed, the corresponding data is changed to pending review status, and S709 is executed.

[0153] Otherwise if If the value is less than 0.7, it is determined to be an abnormal timestamp and timestamp correction is adopted: Since the edge nodes in the same area use a unified periodic trigger for acquisition, the standard discrete acquisition timestamp recorded by the neighboring edge nodes with normal time synchronization within the same acquisition period is directly used as the correction timestamp of the local measurement data to be verified (i.e., correction timestamp = neighboring standard acquisition timestamp) to eliminate the rolling drift of the local clock. Otherwise if <0.6, and It has valid calculation results and switch-current on / off verification results. =1. Upstream and downstream voltage gradient verification results =1. Conservation verification result of the total score table of the transformer area =1. Time reliability factor Link Trust Factor If the deviation is ≥0.6, it is determined to be an abnormal local measurement deviation, and compensation is made using neighboring measurement data of the same type.

[0154] Edge nodes extract the expected value of each neighboring node in the effective neighboring candidate set, and then perform a weighted average based on the neighboring mutual verification weights of each neighboring node to obtain the neighborhood expected value. .

[0155] Let the local abnormal measurement value be The corrected local measurement data is Then, it should be corrected according to the following formula: ; in, The closer it is to 0.6, the less the deviation between local measurement data and neighboring data, and the higher the weight of local outliers; The closer it is to 0, the greater the deviation between the local measurement data and the neighboring data, and the higher the weight of the neighborhood conversion expectation value.

[0156] When neighborhood mutual verification factor If the calculation is not possible, the effective neighborhood candidate set is empty, or the expected value of the effective neighborhood cannot be obtained, the automatic compensation of neighborhood data will not be performed. Instead, the business access status of the corresponding data will be changed to pending review, and the relevant data will be uploaded to the main station in accordance with S709.

[0157] Otherwise, if the result of the conservation check of the total and sub-tables of the distribution area is... =0, and the switch-current continuity test result =1. Upstream and downstream voltage gradient verification results =1, Neighborhood Mutual Verification Factor Time reliability factor Link Trust Factor If the value is ≥0.6, it is determined to be an anomaly in the conservation of the total score table of the district.

[0158] The edge node first verifies whether the electricity data of the main meter of the transformer area and the electricity data of its upstream feeder outlet metering point in the same settlement period meet the preset electricity gradient conservation relationship, and verifies whether the line loss rate parameter used in this verification is valid, and whether the settlement period corresponding to the main meter data of the transformer area and the data of each sub-meter is consistent.

[0159] When the total electricity data of the transformer area passes the cross-verification, the line loss rate parameter is valid, the total data of the transformer area corresponds to the same settlement period as the data of each sub-meter, and only one sub-meter in all sub-meters fails the data quality verification while the data of the remaining sub-meters passes the data quality verification, the sub-meter that fails the data quality verification is identified as the target abnormal sub-meter.

[0160] Let the total electricity consumption of the transformer substation be... The estimated line loss is The sum of the electricity consumption of all normal sub-meters except the target abnormal sub-meter is Then, the corrected electricity level for the target abnormal sub-meter is calculated according to the following formula. : .

[0161] Edge nodes to correct power Replace the original power data of the target abnormal sub-table, and use the replaced data as the repaired data. Then, proceed to S705 to S706 to perform the comprehensive quality reliability calculation and access condition review after repair.

[0162] When the total electricity consumption data of the transformer substation fails cross-verification with the upstream feeder outlet metering data, the line loss rate parameter is invalid, the settlement cycle of the total electricity consumption data of the transformer substation is inconsistent with that of each sub-meter, there are two or more abnormal sub-meters at the same time, or the target abnormal sub-meter cannot be uniquely identified, automatic reverse calculation correction will not be performed. Instead, the business access status of the corresponding data will be changed to pending review, and the total electricity consumption data of the transformer substation, all sub-meter data, line loss rate parameters, settlement cycle information, and the already obtained data will be uploaded to the main station in accordance with S709. , , , Factor values.

[0163] Otherwise, if the link trust factor <0.6, and the switch-current continuity test result =1. Upstream and downstream voltage gradient verification results =1. Conservation verification result of the total score table of the distribution area =1, Neighborhood Mutual Verification Factor ≥0.6, Time Reliability Factor If the value is ≥0.7, it is determined to be an abnormal link transmission.

[0164] The edge node reads the link connectivity status, retransmission records, packet loss records, and link switching records corresponding to the measurement data to be verified, and selects the repair method in the following order: (1) If the link record indicates that the data transmitted in the first transmission is incomplete or there is a verification anomaly, and there is a complete retransmission success record corresponding to the same acquisition period, then the data received after the last complete retransmission success will replace the data received in the first transmission as the corrected local measurement data. (2) If there is no complete retransmission success record, but there is a link switching record, and the complete measurement data corresponding to the same acquisition period is successfully received after the link switching, then the data received before the link switching is replaced with the complete measurement data received after the link switching, and used as the corrected local measurement data. (3) If there is no complete retransmission successful data or complete data received after link switching, but there is an effective neighborhood conversion expectation value obtained according to S504, the effective neighborhood conversion expectation value is used to compensate the local measurement data. (4) If there is no complete retransmission successful data, complete data received after link switching, and effective neighborhood conversion expectation value, then automatic repair will not be performed. The service access status of the data will be changed to pending review status, and the original measurement data and comprehensive quality reliability will be uploaded to the main station in accordance with S709. as well as , , and Each factor value is accompanied by a "link abnormal and cannot be repaired locally" flag; (5) If the communication link is continuously disconnected, resulting in no valid measurement value being received in this acquisition cycle, a data missing mark is set, and the link transmission abnormality branch is no longer processed. Instead, the data missing status judgment branch described in S702 is entered. If there is valid compensation data, the missing measurement compensation is performed. If there is no valid compensation data, the status is changed to pending review.

[0165] Only if there are no missing data markers and the switch-current continuity test results are available. =1. Upstream and downstream voltage gradient verification results =1. Conservation verification result of the total score table of the transformer area =1. Only when the time reliability factor TF≥0.7, the neighborhood mutual verification factor NF≥0.6, and the link reliability factor LF≥0.6 are all components considered to be in normal condition.

[0166] When the data is judged by its overall quality and credibility When the system enters a repairable state and meets the normal conditions of each component mentioned above, this repair is a no-operation. The edge nodes do not modify the data values ​​or data acquisition timestamps. The original measurement data obtained in S701 is used as the repaired data, and the system proceeds to S705 to S706 to perform a post-repair verification.

[0167] S705-S706: Post-repair review. Edge nodes re-execute S102-S106 on the repaired data to recalculate the overall quality reliability. .like If the target service access threshold is ≥ (0.80 in this embodiment, which can be configured separately according to the specific needs of downstream services), the repaired data will be provided to the target service (such as line loss analysis, load forecasting, lean operation and maintenance of transformer areas, etc.); if T' still does not meet the target service access threshold, it will be switched to the pending review state, and S709 will be executed to upload the data before and after repair and the five values ​​to the main station.

[0168] S707~S709: Handling of data in repair success, isolation status, and pending review status: For data determined to be in an isolated status, the edge node directly executes local blocking (S708) without attempting repair; for data determined to be in a pending review status, the edge node executes S709, uploading the original measurement data and... , , , , .

[0169] Example 2: A data quality control method for distribution networks based on edge computing includes: edge nodes acquiring locally collected measurement data to be verified, switch status data, communication link status data, and time synchronization status data; performing consistency verification on the measurement data to be verified based on pre-configured electrical consistency constraint rules to obtain an electrical consistency factor; the electrical consistency constraint rules include switch-current on / off relationship, upstream and downstream voltage gradient relationship, and the conservation relationship between the main and sub-meters in the distribution area; Edge nodes determine the time reliability factor based on the time synchronization status, and determine the neighborhood mutual verification time window based on the time reliability factor; within the neighborhood mutual verification time window, they acquire similar measurement data collected by neighboring edge nodes in the same area through nearby communication on the edge side, and compare the measurement data to be verified with the similar measurement data based on the neighborhood mutual verification weight to obtain the neighborhood mutual verification factor; the edge nodes determine the link reliability factor based on the communication link status of the communication link corresponding to the measurement data to be verified. Edge nodes integrate four factors: electrical consistency factor, neighborhood mutual verification factor, time reliability factor, and link reliability factor, to obtain a comprehensive quality reliability score. Based on the comprehensive quality reliability score, the business access status is determined, which includes repairable status, isolated status, and pending review status. For data in a repairable state, perform local repair; for data in an isolated state, perform local blocking; for data in a pending review state, upload the original measurement data of the data to be reviewed, as well as the corresponding comprehensive quality reliability and the above four factors, to the main site.

[0170] Furthermore, the measurement data to be verified includes voltage measurement data and meter data; the meter data includes the total meter data and sub-meter data for the distribution area; the configuration process of the electrical consistency constraint rules includes: the edge node obtains the topology version identifier and topology connection relationship of the distribution network, determines the feeder position, distribution area position, switch position and meter connection position corresponding to the edge node, and generates the corresponding electrical consistency constraint rules for the topology position based on the feeder position, distribution area position, switch position and meter connection position.

[0171] Furthermore, the generation process of the electrical consistency factor includes: edge nodes determining whether the current on / off state of the circuit associated with the switch matches the switch status data to obtain the switch-current on / off verification result; determining whether the voltage change direction conforms to the topology power supply direction based on the voltage measurement data of upstream and downstream nodes to obtain the upstream and downstream voltage gradient verification result; determining whether the power conservation relationship meets the preset conservation conditions based on the total meter data and sub-meter data of the transformer area to obtain the transformer area total and sub-meter conservation verification result; and edge nodes integrating the switch-current on / off verification result, the upstream and downstream voltage gradient verification result, and the transformer area total and sub-meter conservation verification result to obtain the electrical consistency factor.

[0172] Furthermore, the generation process of the time reliability factor includes: the edge node acquiring the local clock state, the time source state, the most recent time transmission record and the data acquisition timestamp, determining the time reliability factor based on the degree of deviation between the local clock and the time source, and adjusting the neighborhood mutual verification time window based on the time reliability factor.

[0173] Furthermore, the process of determining the neighborhood mutual verification time window includes: when the time credibility factor indicates that the time synchronization state is credible, the edge node uses a first time window for neighborhood mutual verification; when the time credibility factor indicates that the time synchronization state is abnormal, the edge node uses a second time window for neighborhood mutual verification; wherein, the time range covered by the second time window is greater than the time range covered by the first time window.

[0174] Furthermore, the generation process of the neighborhood mutual verification factor includes: edge nodes determining the measurement type and topology location corresponding to the measurement data to be verified; screening neighboring edge nodes that have an electrical relationship with the topology location within the same transformer area; acquiring the same type of measurement data collected by the neighboring edge nodes within the neighborhood mutual verification time window through edge-side proximity communication; comparing the deviation and trend of the measurement data to be verified with the same type of measurement data; and generating the neighborhood mutual verification factor based on the deviation comparison results and trend comparison results.

[0175] Furthermore, the process of determining the neighborhood mutual verification weight includes: the edge node calculates the initial weight corresponding to each neighborhood edge node through a preset weight calculation function based on the electrical distance, communication delay, and data acquisition time difference between the neighborhood edge node and the corresponding measurement point of the measurement data to be verified, and normalizes each initial weight to obtain the neighborhood mutual verification weight; wherein, the initial weight is negatively correlated with the electrical distance, the communication delay, and the data acquisition time difference.

[0176] Furthermore, the process of generating the link trust factor includes: the edge node reading the communication link record that the measurement data to be verified passes through from the acquisition terminal to the edge node, the communication link record including link connectivity status, transmission delay, retransmission record, packet loss record and link switching record, and the edge node determining the link trust factor based on the communication link record.

[0177] Furthermore, the process of determining the service access status includes: edge nodes fusing electrical consistency factors, neighborhood mutual verification factors, time reliability factors, and the link reliability factors according to preset fusion rules to obtain a comprehensive quality reliability; when the comprehensive quality reliability meets the local repair access conditions, the service access status is determined to be a repairable status; when the comprehensive quality reliability meets the service blocking conditions, the service access status is determined to be an isolated status; when the comprehensive quality reliability does not meet the local repair access conditions and does not meet the service blocking conditions, the service access status is determined to be a pending review status.

[0178] Furthermore, the process of local repair for data in a repairable state includes: the edge node selects the corresponding repair method according to the anomaly type of the measurement data to be verified, the repair method includes timestamp correction, adjacent acquisition cycle data compensation, neighboring similar type measurement data compensation, and data correction based on the electrical consistency constraint rules; the edge node recalculates the comprehensive quality reliability of the repaired data, and provides the repaired data to the target service when the recalculated comprehensive quality reliability meets the target service access conditions.

[0179] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for data quality control in distribution networks based on edge computing, characterized in that, include: Edge nodes acquire locally collected measurement data to be verified, switch status data, communication link status data, and time synchronization status data; Based on pre-configured electrical consistency constraint rules, the consistency of the measurement data to be verified is checked to obtain the electrical consistency factor; Electrical consistency constraints include switch-current on / off relationships, upstream and downstream voltage gradient relationships, and the conservation relationship between the main and sub-meters in the distribution area; Edge nodes determine the time reliability factor based on the time synchronization status, and determine the neighborhood mutual verification time window based on the time reliability factor; within the neighborhood mutual verification time window, they acquire similar measurement data collected by neighboring edge nodes in the same area through nearby communication on the edge side, and compare the measurement data to be verified with the similar measurement data based on the neighborhood mutual verification weight to obtain the neighborhood mutual verification factor; the edge nodes determine the link reliability factor based on the communication link status of the communication link corresponding to the measurement data to be verified. Edge nodes integrate four factors: electrical consistency factor, neighborhood mutual verification factor, time reliability factor, and link reliability factor, to obtain a comprehensive quality reliability score. Based on the comprehensive quality reliability score, the business access status is determined, which includes repairable status, isolated status, and pending review status. Perform local repair on data that is in a repairable state, and perform local blocking on data that is in an isolated state; For data that is pending review, upload the original measurement data of the data to be reviewed, along with the corresponding comprehensive quality reliability and the four factors mentioned above, to the main site.

2. The method for data quality control of distribution networks based on edge computing according to claim 1, characterized in that: The measurement data to be verified includes voltage measurement data and meter data; the meter data includes the total meter data for the transformer area and the sub-meter data. The configuration process of electrical consistency constraint rules includes: edge nodes obtaining the topology version identifier and topology connection relationship of the distribution network, determining the feeder location, transformer location, switch location and meter connection location corresponding to the edge node, and generating electrical consistency constraint rules for the corresponding topology location based on the feeder location, transformer location, switch location and meter connection location.

3. The data quality control method for distribution networks based on edge computing according to claim 2, characterized in that: The generation process of the electrical consistency factor includes: edge nodes determining whether the current on / off state of the circuit associated with the switch matches the switch status data to obtain the switch-current on / off verification result; determining whether the voltage change direction conforms to the power supply direction of the topology based on the voltage measurement data of upstream and downstream nodes to obtain the upstream and downstream voltage gradient verification result; determining whether the energy conservation relationship meets the preset conservation conditions based on the total meter data and sub-meter data of the transformer area to obtain the transformer area total and sub-meter conservation verification result; and edge nodes integrating the switch-current on / off verification result, the upstream and downstream voltage gradient verification result, and the transformer area total and sub-meter conservation verification result to obtain the electrical consistency factor.

4. The data quality control method for distribution networks based on edge computing according to claim 1, characterized in that: The generation process of the time reliability factor includes: the edge node acquiring the local clock status, the time source status, the most recent time synchronization record and the data acquisition timestamp, determining the time reliability factor based on the degree of deviation between the local clock and the time source, and adjusting the neighborhood mutual verification time window based on the time reliability factor.

5. The data quality control method for distribution networks based on edge computing according to claim 4, characterized in that: The process of determining the neighborhood mutual verification time window includes: when the time credibility factor indicates that the time synchronization state is credible, the edge node uses the first time window for neighborhood mutual verification; when the time credibility factor indicates that the time synchronization state is abnormal, the edge node uses the second time window for neighborhood mutual verification; wherein, the time range covered by the second time window is greater than the time range covered by the first time window.

6. The data quality control method for distribution networks based on edge computing according to claim 5, characterized in that: The generation process of the neighborhood mutual verification factor includes: edge nodes determining the measurement type and topology location corresponding to the measurement data to be verified; screening neighboring edge nodes with electrical correlation to the topology location within the same transformer area; acquiring the same type of measurement data collected by the neighboring edge nodes within the neighborhood mutual verification time window through edge-side proximity communication; comparing the deviation and trend of the measurement data to be verified with the same type of measurement data; and generating the neighborhood mutual verification factor based on the deviation comparison results and trend comparison results.

7. The data quality control method for distribution networks based on edge computing according to claim 6, characterized in that: The process of determining the neighborhood mutual verification weight includes: the edge node calculates the initial weight corresponding to each neighborhood edge node according to the electrical distance, communication delay and data acquisition time difference between the neighborhood edge node and the corresponding measurement point of the measurement data to be verified, through a preset weight calculation function, and normalizes each initial weight to obtain the neighborhood mutual verification weight; wherein, the initial weight is negatively correlated with the electrical distance, the communication delay and the data acquisition time difference.

8. The data quality control method for distribution networks based on edge computing according to claim 1, characterized in that: The process of generating the link trust factor includes: the edge node reads the communication link record that the measurement data to be verified is transmitted from the acquisition terminal to the edge node, the communication link record includes the link connectivity status, transmission delay, retransmission record, packet loss record and link switching record, and the edge node determines the link trust factor based on the communication link record.

9. The data quality control method for distribution networks based on edge computing according to claim 8, characterized in that: The process of determining the service access status includes: edge nodes fusing electrical consistency factors, neighborhood mutual verification factors, time reliability factors, and link reliability factors according to preset fusion rules to obtain a comprehensive quality reliability; when the comprehensive quality reliability meets the local repair access conditions, the service access status is determined to be a repairable status; when the comprehensive quality reliability meets the service blocking conditions, the service access status is determined to be an isolated status; when the comprehensive quality reliability does not meet the local repair access conditions and does not meet the service blocking conditions, the service access status is determined to be a pending review status.

10. A method for controlling the quality of distribution network data based on edge computing according to claim 9, characterized in that: The process of local repair for data in a repairable state includes: edge nodes selecting the corresponding repair method according to the anomaly type of the measurement data to be verified, the repair method including timestamp correction, data compensation for adjacent acquisition cycles, compensation for similar measurement data in the neighboring area, and data correction based on the electrical consistency constraint rules; the edge nodes recalculate the comprehensive quality reliability of the repaired data, and provide the repaired data to the target service when the recalculated comprehensive quality reliability meets the target service access conditions.

Images