Research On Catenary Equipment Defect Identification Based On Deep Learning

High-speed railway catenary power supply system is a complex system,containing a large number of parts,different types,each play different functions,to ensure that high-speed bullet train can get power normally,the failure of parts will affect the operation safety of high-speed railway.At present,the maintenance of catenary equipment of high-speed railway mainly relies on manual visual inspection to check whether the equipment has defects or potential safety hazards.The maintenance speed is slow,and the maintenance is carried out annually according to the maintenance cycle of the equipment,so the defects of the equipment are not easy to be found in time.In order to solve the above problems,each road bureau has introduced 6C system to monitor the operation state of catenary equipment,while 4C system mainly detects the state of contact suspension.With a large amount of picture data,all rely on manual analysis of the picture to find the defects of equipment.The labor intensity of personnel is very high,and the labor efficiency is not high.In order to solve the problem of labor intensity and improve labor efficiency,we can consider using scientific and technological means to solve.At present,the image processing technology has developed rapidly,the application in various fields,and has obtained the remarkable effect,so in this paper,based on the convolutional neural network image recognition technology combined with catenary equipment defect image scene,to test the state of high speed railway catenary device monitoring,realize the equipment safety hidden danger of illness,and fixing the defective equipment.This article main research convolution neural network catenary equipment parts and components defect recognition methods and models,made some achievements,the first target detection module based on regional framework,design contains a lightweight network,global attention module of feature extraction and classifier and detector of three mutually reinforcing some parts target detection network,finally through the model training and validation,The effectiveness,accuracy and efficiency of the catenary parts defect identification method were tested,which can assist operators to improve the analysis efficiency of image data,reduce the intensity of manual labor,and ensure that the catenary equipment defects can be found and dealt with in time,eliminate potential safety hazards and ensure the safety of high-speed railway operation.