A rail damage detection and measurement system using neural networks

Abstract

Rail defects and damages often cause train accidents. In this paper, an onboard measurement system for measuring the rail-robot's excursions from the rails' midlines and the rail-robot's heights above the rails is presented. In this system, two groups of proximity transducers are placed above the two parallel rail tracks. This measurement system is an important part of a comprehensive online rail damages detection, measurement and reparation system, which is called the rail-robot. To deal with the nonlinearity of the measurement models, the coupling between the outputs, and the noise contamination, a neural network method is proposed for building high precision measurement models. Moreover, different measurement models for different types of rail tracks are also built based on the proposed neural network module. Experimental results show that this neural network based measurement system has high precision and is suitable for online rail damage detection and measurement applications.