CN-121978469-A - Power distribution network short-time grounding reason identification method based on self-adaptive weighting K nearest neighbor

CN121978469ACN 121978469 ACN121978469 ACN 121978469ACN-121978469-A

Abstract

The invention discloses a power distribution network short-time grounding reason identification method based on self-adaptive weighting K nearest neighbor. The method comprises the steps of determining the number of target neighbors based on a plurality of preset standard sample waveforms and actual trigger reasons of the standard sample waveforms, collecting original transient recording data corresponding to a short-time grounding event of a current power distribution network, extracting current waveforms based on the original transient recording data, calculating similarity distances between the current waveforms and the standard sample waveforms respectively to obtain the similarity distances corresponding to the standard sample waveforms, screening out the target neighbor waveforms matched with the number of the target neighbors based on the similarity distances corresponding to the standard sample waveforms, and determining the trigger reasons corresponding to the short-time grounding event of the current power distribution network based on the trigger reasons corresponding to the target neighbor waveforms. The invention solves the technical problem that the identification accuracy is low because the identification of the short-time grounding reasons at present mostly depends on manual experience.

Inventors

YAO YUHAI
ZHAO QIAO
CONG ZIHAN
LIU RUOXI
SHI QINGXIN
LIU WENXIA
DAI HANQI
WANG DAWEI
WANG CUNPING

Assignees

国网北京市电力公司
华北电力大学

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. A power distribution network short-time grounding reason identification method based on self-adaptive weighting K nearest neighbor is characterized by comprising the following steps: determining the number of target neighbors based on a plurality of preset standard sample waveforms and actual trigger reasons of the plurality of standard sample waveforms, wherein the plurality of standard sample waveforms are waveforms generated by short-time grounding signals of a power distribution network; Collecting original transient recording data corresponding to a short-time grounding event of a current power distribution network; extracting a current waveform based on the original transient recording data; Calculating similarity distances between the current waveform and the plurality of standard sample waveforms respectively to obtain similarity distances corresponding to the plurality of standard sample waveforms respectively; screening out a plurality of target neighbor waveforms matched with the target neighbor number based on the similarity distances corresponding to the standard sample waveforms; And determining the trigger reason corresponding to the short-time grounding event of the current power distribution network based on the trigger reason corresponding to each of the plurality of target neighbor waveforms.
2. The method of claim 1, wherein the determining the target number of neighbors based on the preset plurality of standard sample waveforms and the actual trigger causes of each of the plurality of standard sample waveforms comprises: Acquiring a plurality of candidate neighbor numbers; Based on a plurality of preset standard sample waveforms and actual trigger reasons of the plurality of standard sample waveforms, respectively calculating trigger reason prediction accuracy corresponding to the number of the plurality of candidate neighbors; And selecting the number of candidate neighbors with the trigger reason prediction accuracy exceeding a preset threshold as the number of target neighbors.
3. The method according to claim 2, wherein calculating the trigger cause prediction accuracy corresponding to a target candidate neighbor number based on a preset plurality of standard sample waveforms and actual trigger causes of each of the plurality of standard sample waveforms, wherein the target candidate neighbor number is any one of the plurality of candidate neighbor numbers, includes: dividing the plurality of standard sample waveforms into a plurality of waveform subsets of a preset number; determining a test waveform subset and a training waveform subset based on the plurality of waveform subsets; Calculating similarity distances between a plurality of standard sample waveforms in the test waveform subset and standard sample waveforms in the training waveform subset respectively; Selecting standard sample waveforms with the number matched with the target candidate neighbor number from a plurality of similarity distances corresponding to a plurality of standard sample waveforms in the test waveform subset, and taking the standard sample waveforms as sample neighbor waveforms corresponding to the standard sample waveforms in the test waveform subset; Determining a predicted trigger reason corresponding to each of a plurality of standard sample waveforms in the test waveform subset based on a trigger reason corresponding to a sample neighbor waveform corresponding to each of a plurality of standard sample waveforms in the test waveform subset; and determining the trigger reason prediction accuracy corresponding to the target candidate neighbor number based on the predicted trigger reasons corresponding to the standard sample waveforms in the test waveform subset and the actual trigger reasons corresponding to the standard sample waveforms in the test waveform subset.
4. The method of claim 1, wherein the extracting a current waveform based on the original transient recording data comprises: Analyzing the time sequence data of the three-phase current from the original transient recording data; and based on the time sequence data of the three-phase current, carrying out point-by-point addition on a plurality of preset sampling points, and determining the current waveform.
5. The method of claim 1, wherein the determining the trigger cause corresponding to the current power distribution network short-time ground event based on the trigger causes corresponding to each of the plurality of target neighbor waveforms comprises: determining decision weights corresponding to the target neighbor waveforms based on the similarity distance values corresponding to the target neighbor waveforms; Determining a plurality of candidate trigger reasons based on the actual trigger reasons corresponding to the target neighbor waveforms respectively; Determining total decision weights corresponding to the candidate trigger reasons based on the decision weights corresponding to the target neighbor waveforms; determining confidence scores corresponding to the plurality of candidate trigger reasons based on the total decision weights corresponding to the plurality of candidate trigger reasons; And selecting the candidate triggering reasons with confidence scores meeting preset conditions as triggering reasons corresponding to the short-time grounding event of the current power distribution network.
6. The method of claim 5, wherein the determining the confidence score for each of the plurality of candidate trigger causes based on the total decision weights for each of the plurality of candidate trigger causes comprises: determining confidence scores corresponding to the plurality of candidate trigger reasons based on total decision weights corresponding to the plurality of candidate trigger reasons according to a preset formula, wherein the preset formula is as follows: Wherein, the For candidate trigger reasons The confidence score of the corresponding one of the two, For candidate trigger reasons Corresponding overall decision weights.
7. The utility model provides a distribution network short-time ground connection reason discernment device based on self-adaptation weighting K is close, its characterized in that includes: the first determining module is used for determining the number of target neighbors based on a plurality of preset standard sample waveforms and actual trigger reasons of the standard sample waveforms, wherein the standard sample waveforms are waveforms generated by short-time grounding signals of the power distribution network; The acquisition module is used for acquiring original transient recording data corresponding to the short-time grounding event of the current power distribution network; the extraction module is used for extracting a current waveform based on the original transient recording data; The calculation module is used for calculating the similarity distances between the current waveforms and the standard sample waveforms respectively to obtain the similarity distances corresponding to the standard sample waveforms respectively; the screening module is used for screening a plurality of target neighbor waveforms matched with the target neighbor number based on the similarity distances corresponding to the standard sample waveforms; and the second determining module is used for determining the triggering reasons corresponding to the short-time grounding event of the current power distribution network based on the triggering reasons corresponding to the target neighbor waveforms.
8. A non-volatile storage medium, characterized in that the non-volatile storage medium comprises a stored program, wherein the device in which the non-volatile storage medium is controlled to execute the method for identifying the short-term grounding cause of the power distribution network based on the adaptive weighted K nearest neighbor according to any one of claims 1 to 6 when the program runs.
9. A computer device is characterized by comprising a memory and a processor, The memory stores a computer program; The processor is configured to execute a computer program stored in the memory, where the computer program when executed causes the processor to execute the method for identifying a short-time grounding reason of a power distribution network based on an adaptively weighted K nearest neighbor according to any one of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the method for identifying short-term ground causes of a power distribution network based on adaptively weighted K-nearest neighbors according to any one of claims 1 to 6.

Description

Power distribution network short-time grounding reason identification method based on self-adaptive weighting K nearest neighbor Technical Field The invention relates to the technical field of power distribution network detection, in particular to a power distribution network short-time grounding reason identification method based on self-adaptive weighting K nearest neighbor. Background Triggering reason identification for instantaneous grounding events of a power distribution network mainly depends on an artificial intelligence method, in particular to a deep learning technology. However, the application of the deep learning method in the identification of the trigger reason of the instantaneous grounding signal of the power distribution network faces significant technical challenges and limitations. These challenges mainly stem from the dilemma of "high quality small samples", namely that although the in-situ recording type terminal is able to capture massive amounts of transient event data, the number of high quality samples that can be obtained for a truly "well-defined, label accurate" is extremely limited due to the transient nature of the event and the high cost of in-situ verification. This directly results in the deep learning model failing to learn fully the diversified features during training, thereby affecting the generalization performance and recognition accuracy of the model. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the invention provides a method for identifying short-time grounding reasons of a power distribution network based on self-adaptive weighted K nearest neighbor, which at least solves the technical problem that identification accuracy is low because identification of the current short-time grounding reasons is mostly dependent on manual experience. According to one aspect of the embodiment of the invention, a power distribution network short-time grounding reason identification method based on self-adaptive weighting K nearest neighbor is provided, and the method comprises the steps of determining the number of target nearest neighbor based on a plurality of preset standard sample waveforms and actual triggering reasons of the standard sample waveforms, wherein the standard sample waveforms are waveforms generated by power distribution network short-time grounding signals, collecting original transient state recording data corresponding to a current power distribution network short-time grounding event, extracting current waveforms based on the original transient state recording data, calculating similarity distances between the current waveforms and the standard sample waveforms respectively to obtain similarity distances corresponding to the standard sample waveforms, screening out the target nearest neighbor waveforms matched with the number of the target nearest neighbor based on the similarity distances corresponding to the standard sample waveforms, and determining triggering reasons corresponding to the current short-time power distribution network grounding event based on triggering reasons corresponding to the target nearest neighbor waveforms. Optionally, determining the target neighbor number based on a preset plurality of standard sample waveforms and actual trigger reasons of the plurality of standard sample waveforms respectively comprises obtaining a plurality of candidate neighbor numbers, calculating trigger reason prediction accuracy corresponding to the plurality of candidate neighbor numbers respectively based on the preset plurality of standard sample waveforms and the actual trigger reasons of the plurality of standard sample waveforms respectively, and selecting the candidate neighbor number with the trigger reason prediction accuracy exceeding a preset threshold as the target neighbor number. The method comprises the steps of selecting a standard sample waveform with the number matched with the number of target candidate neighbors from a plurality of similarity distances respectively corresponding to the plurality of standard sample waveforms in a test waveform subset, taking the standard sample waveform with the number matched with the number of target candidate neighbors as a sample neighbor waveform respectively corresponding to the plurality of standard sample waveforms in the test waveform subset, determining a predicted trigger factor respectively corresponding to the plurality of standard sample waveforms in the test waveform subset based on the plurality of waveform subsets, and determining the actual trigger factor of the target candidate neighbor based on the predicted trigger factor respectively corresponding to the plurality of standard sample waveforms in the test waveform subset. Optionally, extracting the current waveform based on the original transient recording data comprises the steps of analyzing time sequence data of the three-phase current from the orig