Method for automatically detecting fault of stay cable of overhead line system based on image enhancement processing

Abstract

The invention discloses a method for automatically detecting faults of a stay cable of an overhead line system based on image enhancement processing. The method specifically comprises the steps: firstly, carrying out the enhancement of an overhead line system image through a 3D DLUT deep convolutional neural network, and enabling the image to be clearer and easier to recognize; secondly, using the angle-based CNN network DAL for positioning a stay cable, and introducing angle parameters into a target detection network, so that the stay cable is positioned more accurately, and subsequent fault analysis is more facilitated; and finally, detecting the stay cable by utilizing Hough transform, and proposing a curvature change-based stay cable loosening defect criterion based on peak distribution of Hough transform. According to the method, the overhead line system image can be effectively enhanced, the stay cable can be accurately positioned, and all defects of a stay cable component under different conditions can be accurately positioned.

Description

Technical field[0001]The present invention belongs to the field of high-speed rail image intelligent detection techniques, and more particularly to an automatic detection method of contact mesh wiring based on image enhancement processing.Background technique[0002]The contact network is one of the most important equipment in the high-speed rail system, responsible for the power supply of the high-speed train, such asimage 3 Indicated. There are many components in the contact network support, such as insulators, support, cable, locator base, and the like. These components are assembled together to secure the contact line, so that the train can stably receive current. Once one of the components fails, the safety of the entire train will be seriously threatened.[0003]Currently, there are many ways to depthly convilient neural network (CNNS) and image processing for different contact network components, and there is also a certain effect. Han Wei proposed a detection and positioning met...

AI Technical Summary

Problems solved by technology

[0004] In the cable-stayed cable fault detection method proposed by Liu Wenqiang, although it achieves the purpose of cable-stayed component fault detection, there are some serious shortcomings in some cases. 1) If there is only one catenary cable base and cable-stayed positioning hook , the diagonal line will not be positioned to

This will result in undetectable failure of the cable-stayed

2) In one picture, there may be multiple catenary cable bases and stay wire positioning hooks, which cannot match each other

This can lead to misalignment of the stay wires and failure to detect failures of the stay wire components

3) Because the supplementary light of the image acquisition system may not completely illuminate the oblique line, and the light is too dark, making it difficult to extract the oblique line

Images