DETECTION OF RAILWAY TRACK FAILURES USING NEURAL NETWORK

Author (s) : V.Meena, M.Preethi, T.Manikumar, Dr.K.Poongothai

---

Abstract :
This paper suggest a railway fracture detectingsystem. This studyoffers a classificationsystemthat use deep learningand convolutional neural networks to categorise every fracture in railway rails (CNNs).In the railway system In terms of railway track unidentified fracture in rail tracks in Indian railway, accidents are the main source ofconcern. The majorityofaccidents occur when railway tracks crack, resulting in the loss of precious lives and economic loss. Using a crack classification system, it is necessary to monitor the track's health on a regular basis. This research uses image processing technologies to classify faults on railway tracks, preventing train derailment. The goal is to create a tracks fragment detection system that combines CNNs with image pre- processingtechniques.It showed how cnn models (CNNs) maybe usedtoblend image processing efficiency with deep learning has been very successful in determining whether or not a railway track crack has occurred. Using a CNN trained with a publicly available image dataset, a number of neuron- and layer- wise visualisation methods were employed. As a result, neural networks have been reported to be able to record the colours and textures of lesions related to correspondingrailwaytrack breaks, whichis similarto humandecision-making.

Keywords : Fragment detection system, CNN, visualisation methods, neural networks.

Conference Date : 13/11/2020

Pages : 48

Paper ID : 20048