Research On Detection Technology Of Key Components Of Catenary Suspension Based On Deep Learning

With the continuous development of my country’s high-speed railway transportation,high-speed railway transportation has become the main mode of transportation in my country.The catenary suspension device is the main power supply device for high-speed trains.Any failure may cause train power outage,or serious traffic accidents.Casualties,and the traditional method of manually inspecting the catenary is inefficient and unstable.In this regard,this paper proposes a fault detection method based on deep learning algorithm to detect the faults of two key components in the catenary suspension of high-speed rail and form experiments to verify the accuracy of the algorithm in this paper.The main research contents and conclusions of this article on the key components of the high-speed rail catenary are divided into the following points:1.HND-Net target detection algorithm is established for the two small key components of nut and split pin in this paper to realize the preliminary positioning of the two key components.This paper proposes two main modules,the residual feature enhancement layer and the cross-pyramid adaptive fusion layer,in the target detection algorithm.The effectiveness of this module is proved through ablation experiments,and the target location of the nut and split pin area in this paper is carried out to verify the accuracy of the algorithm in this paper.Sex.2.Aiming at detecting the fault of U-shaped hoop nut,a fault detection algorithm based on instance segmentation algorithm is proposed.By analyzing the two faults of the hoop nut image obtained by the initial positioning,the missing top nut and the loosening of the top nut,a method for image segmentation of the four nuts using the instance segmentation algorithm is proposed,and finally the segmentation image is used to set the ratio threshold value Loose fault diagnosis,and corresponding experiments have been carried out to verify the reliability of the algorithm in this paper.3.Aiming at the cotter pin fault in the hoop area,a fault detection algorithm based on the key point detection algorithm is proposed.By analyzing the missing and insufficient opening angle faults of the cotter pin image in the hoop area obtained by the initial positioning,a method of detecting the three key points of the cotter pin using a key point detection algorithm is proposed,and finally the key points are used to detect the key points obtained from the image.Point setting angle threshold realizes the fault diagnosis of insufficient opening angle of cotter pin,and corresponding experiments have been carried out to verify the reliability of the algorithm in this paper.