Article: Railway track fault detection using optimised convolution neural network Journal: International Journal of Reliability and Safety (IJRS) 2024 Vol.18 No.2 pp.163 - 186 Abstract: Railway accidents are an under-scrutinised cause of death in India. Train accidents are caused by various consequences of collisions, derailments, signal errors and so on. Furthermore, when train derailments become disastrous, they can have tremendous repercussions. It is practically difficult to identify the cause of the derailment efficiently within a limited period. In recent years, we have been making progress in reducing derailments, but even if not deadly, identifying faulty tracks can waste a lot of time and money. And doing this error-free is a pressing matter, as tracks always experience wear and tear with more usage. Here is where neural networks can pitch in their solution. We can train a model to look at train tracks and identify any issues. This paper goes into the methodology of achieving this and optimising a neural network to predict problems in the track with the best possible accuracy that images can provide. The objective of this paper is to identify, develop and optimise neural networks to detect faulty tracks. In this work, a good Convolution Neural Network model is developed to identify the crack in the railway track. The developed model produced 95.54% accuracy in fault classification. Inderscience Publishers - linking academia, business and industry through research

Title: Railway track fault detection using optimised convolution neural network

Authors: R. Chitra; A.M. Anusha Bamini; D. Brindha; T.M. Chenthil Jegan; S. Stewart Kirubakaran

Addresses: Department of Computer Science and Engineering, Karunya Institute of Technology and Sciences, Coimbatore, Tamil Nadu, India ' Department of Computer Science and Engineering, Karunya Institute of Technology and Sciences, Coimbatore, Tamil Nadu, India ' Department of Computer Science and Engineering, Karunya Institute of Technology and Sciences, Coimbatore, Tamil Nadu, India ' Department of Mechanical Engineering, Sri Ranganathar Institute of Engineering and Technology, Coimbatore, Tamil Nadu, India ' Department of Computer Science and Engineering, Karunya Institute of Technology and Sciences, Coimbatore, Tamil Nadu, India

Abstract: Railway accidents are an under-scrutinised cause of death in India. Train accidents are caused by various consequences of collisions, derailments, signal errors and so on. Furthermore, when train derailments become disastrous, they can have tremendous repercussions. It is practically difficult to identify the cause of the derailment efficiently within a limited period. In recent years, we have been making progress in reducing derailments, but even if not deadly, identifying faulty tracks can waste a lot of time and money. And doing this error-free is a pressing matter, as tracks always experience wear and tear with more usage. Here is where neural networks can pitch in their solution. We can train a model to look at train tracks and identify any issues. This paper goes into the methodology of achieving this and optimising a neural network to predict problems in the track with the best possible accuracy that images can provide. The objective of this paper is to identify, develop and optimise neural networks to detect faulty tracks. In this work, a good Convolution Neural Network model is developed to identify the crack in the railway track. The developed model produced 95.54% accuracy in fault classification.

Keywords: activation function; track; neural network; optimisation; VGG.

DOI: 10.1504/IJRS.2024.139215

International Journal of Reliability and Safety, 2024 Vol.18 No.2, pp.163 - 186

Received: 31 Oct 2023
Accepted: 16 Jan 2024
Published online: 26 Jun 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Railway track fault detection using optimised convolution neural network

Keep up-to-date