CN-121582258-B - Transmission line defect detection method and device and electronic equipment

CN121582258BCN 121582258 BCN121582258 BCN 121582258BCN-121582258-B

Abstract

The invention discloses a method and a device for detecting defects of a power transmission line and electronic equipment. The method comprises the steps of obtaining multi-mode data corresponding to a target power transmission line, retrieving a target architecture model, inputting a power transmission line image into a first part of the target architecture model to obtain an initial defect detection result corresponding to the target power transmission line, and inputting the initial defect detection result and the multi-mode data into a second part of the target architecture model to correct the initial defect detection result and obtain a target defect detection result corresponding to the target power transmission line. The invention solves the technical problem of inaccurate defect detection when the defect detection is carried out on the power transmission line in the related technology.

Inventors

ZHANG HONGYU
CHEN BO
WANG ZHI
CHEN XIAODONG
QI WEIQIANG
AN KANG
FANG XIAO
WU YAO

Assignees

国网北京市电力公司
北京电力经济技术研究院有限公司
中国电力科学研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260127

Claims (10)

1. The method for detecting the defect of the power transmission line is characterized by comprising the following steps of: acquiring multi-mode data corresponding to a target power transmission line, wherein the multi-mode data comprises a power transmission line image; Invoking a target architecture model, wherein the target architecture model comprises a first part and a second part, the target architecture model is trained according to a joint training function and a training data set, the joint training function is used for determining similarity between output of the first part and output of the second part, the difference between the output of the first part and the output of the second part is minimized, the first part is used for defect identification, the second part is used for correcting an output result of the first part, the first part comprises YOLOv and OverLoCK, and the second part comprises LLaVA-MoD; inputting the power transmission line image into a first part of the target architecture model to obtain an initial defect detection result corresponding to the target power transmission line; And inputting the initial defect detection result and the multi-mode data into a second part of the target architecture model to correct the initial defect detection result so as to obtain a target defect detection result corresponding to the target power transmission line.
2. The method of claim 1, wherein prior to invoking the target architecture model, comprising: Determining a training target parameter in the case that the training data set comprises a plurality of sample images, wherein the training target parameter comprises a predetermined similarity threshold; Inputting any one of the plurality of sample images into a first part of the target architecture model to obtain training output corresponding to the first part under the any one of the sample images; determining training output corresponding to the second part under the arbitrary sample image according to the arbitrary sample image and the training output corresponding to the first part under the arbitrary sample image; Determining output similarity between training output corresponding to the first part and training output corresponding to the second part under any sample image according to the joint training function, and determining whether the output similarity is greater than or equal to a determination result of the preset similarity threshold; and stopping training the target architecture model until the determined result is that the output similarity is greater than or equal to the preset similarity threshold.
3. The method of claim 1, wherein prior to invoking the target architecture model, comprising: Under the condition that the training data set comprises positive sample data and negative sample data, the positive sample data is input to the second part to obtain positive sample output corresponding to the positive sample data, wherein the positive sample data is sample data with the defect degree larger than a defect threshold value, and the negative sample data is sample data with the defect degree smaller than or equal to the defect threshold value; inputting the negative sample data into the second part to obtain negative sample output corresponding to the negative sample data; Determining a comparison loss function corresponding to the second portion, wherein the comparison loss function is used for determining the accuracy of defect detection of the second portion; and determining defect detection accuracy corresponding to the second part according to the positive sample output, the negative sample output and the comparison loss function until the defect detection accuracy is greater than or equal to a defect detection accuracy threshold value, and stopping training for the second part.
4. The method of claim 1, wherein inputting the transmission line image into the first portion of the target architecture model to obtain an initial defect detection result corresponding to the target transmission line comprises: determining a defect feature map corresponding to the transmission line image; Dividing the defect feature map to obtain a plurality of subgraphs; Determining sub-defect characteristics corresponding to the plurality of sub-images respectively and a plurality of association characteristics corresponding to the plurality of sub-images respectively, wherein the corresponding plurality of association characteristics represent defect association relations between the corresponding sub-images and other sub-images respectively; And obtaining an initial defect detection result corresponding to the target transmission line according to the sub-defect characteristics and the correlation characteristics respectively corresponding to the sub-graphs.
5. The method of claim 4, wherein obtaining an initial defect detection result corresponding to the target transmission line according to the sub-defect features and the plurality of associated features respectively corresponding to the plurality of sub-graphs comprises: determining an enhancement feature map corresponding to the defect feature map according to the sub-defect features and the multiple associated features respectively corresponding to the multiple sub-graphs; and determining an initial defect detection result corresponding to the target power transmission line according to the enhanced feature map.
6. The method of claim 1, wherein, in the case that the second portion includes a reference defect detection set, inputting the initial defect detection result and the multi-modal data into the second portion in the target architecture model to correct the initial defect detection result to obtain a target defect detection result corresponding to the target power transmission line, includes: determining multi-mode defect characteristics corresponding to the target transmission line according to the multi-mode data; According to the multi-mode defect characteristics, determining a similar power transmission line corresponding to the target power transmission line from a plurality of reference power transmission lines, wherein the similar power transmission line is a reference power transmission line with a difference index of the reference defect characteristics and the multi-mode defect characteristics smaller than a difference threshold value, and the reference defect detection set comprises a plurality of reference power transmission lines, and reference defect characteristics and reference defect detection results respectively corresponding to the plurality of reference power transmission lines; And correcting the initial defect detection result according to the reference defect detection result corresponding to the similar transmission line according to the multi-mode defect characteristics to obtain a target defect detection result corresponding to the target transmission line.
7. The method according to any one of claims 1 to 6, wherein the inputting the transmission line image into the first portion of the target architecture model, after obtaining the initial defect detection result corresponding to the target transmission line, further includes: Determining a target scene parameter corresponding to the target power transmission line under the condition that the target power transmission line is a preset power transmission line and the initial defect detection result comprises a defect area parameter, wherein the preset power transmission line is a power transmission line under a preset scene; Determining and fusing images according to the target scene parameters, the defect area parameters and the power transmission line images; And performing incremental training on the first part according to the fusion image to obtain a first part after incremental training.
8. A transmission line defect detection device, characterized by comprising: The acquisition module is used for acquiring multi-mode data corresponding to the target power transmission line, wherein the multi-mode data comprises a power transmission line image; A retrieving module, configured to retrieve a target architecture model, where the target architecture model includes a first portion and a second portion, the target architecture model is obtained by training according to a joint training function and a training data set, the joint training function is configured to determine a similarity between an output of the first portion and an output of the second portion, minimize a difference between the output of the first portion and the output of the second portion, the first portion is configured to identify a defect, the second portion is configured to correct an output result of the first portion, the first portion includes YOLOv, overLoCK, and the second portion includes LLaVA-MoD; the first determining module is used for inputting the power transmission line image into a first part of the target architecture model to obtain an initial defect detection result corresponding to the target power transmission line; And the second determining module is used for inputting the initial defect detection result and the multi-mode data into a second part of the target architecture model so as to correct the initial defect detection result and obtain a target defect detection result corresponding to the target power transmission line.
9. An electronic device, comprising: A processor; a memory for storing the processor-executable instructions; wherein the processor is configured to execute the instructions to implement the transmission line defect detection method of any one of claims 1 to 7.
10. A computer readable storage medium, characterized in that instructions in the computer readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the transmission line defect detection method according to any one of claims 1 to 7.

Description

Transmission line defect detection method and device and electronic equipment Technical Field The invention relates to the field of defect detection, in particular to a method and a device for detecting defects of a power transmission line and electronic equipment. Background With the continuous increase of power demand and the increasing expansion of grid scale, the operation and maintenance management of power transmission lines becomes more important. Defects in the power transmission line, such as insulator breakage, wire strand breakage, tower corrosion and the like, can be found and processed in time, faults can be effectively prevented, power failure time is reduced, and power supply reliability is improved. However, in the related art, when the defect detection is performed on the power transmission line, the technical problem of inaccurate defect detection exists. 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, a device and electronic equipment for detecting defects of a power transmission line, which at least solve the technical problem of inaccurate defect detection when the defect detection is carried out on the power transmission line in the related technology. According to one aspect of the embodiment of the invention, a power transmission line defect detection method is provided, and the power transmission line defect detection method comprises the steps of obtaining multi-mode data corresponding to a target power transmission line, wherein the multi-mode data comprise a power transmission line image, retrieving a target architecture model, wherein the target architecture model comprises a first part and a second part, the target architecture model is obtained through training according to a combined training function and a training data set, the combined training function is used for determining similarity between output of the first part and output of the second part, the first part is used for defect identification, the second part is used for correcting output results of the first part, the power transmission line image is input into the first part of the target architecture model to obtain an initial defect detection result corresponding to the target power transmission line, and the multi-mode data is input into the second part of the target architecture model to correct the initial defect detection result to obtain a target defect detection result corresponding to the target power transmission line. Optionally, before the target architecture model is invoked, determining training target parameters when the training data set includes a plurality of sample images, wherein the training target parameters include a predetermined similarity threshold, inputting any one of the plurality of sample images to a first portion of the target architecture model to obtain a training output corresponding to the first portion under the any one sample image, determining training output corresponding to the second portion under the any one sample image according to the training output corresponding to the first portion under the any one sample image and the any one sample image, determining output similarity between the training output corresponding to the first portion and the training output corresponding to the second portion under the any one sample image according to the joint training function, and determining whether the output similarity is greater than or equal to a determination result of the predetermined similarity threshold, and stopping training the target architecture model until the determination result is that the output similarity is greater than or equal to the predetermined similarity threshold. Optionally, before the target architecture model is invoked, the method comprises the steps of inputting positive sample data into the second part to obtain positive sample output corresponding to the positive sample data when the training data set comprises positive sample data and negative sample data, wherein the positive sample data is sample data with defect degree larger than a defect threshold value, the negative sample data is sample data with defect degree smaller than or equal to the defect threshold value, inputting the negative sample data into the second part to obtain negative sample output corresponding to the negative sample data, determining a comparison loss function corresponding to the second part, wherein the comparison loss function is used for determining the accuracy degree of defect detection of the second part, determining the defect detection accuracy corresponding to the second part according to the positive sample output until the defect detection accuracy degree is larger than or equal to the defect detection accuracy threshold value, and stopping training the second part. Optionally, the power transmission line image is input to a first part of the target