Railway point mechanisms: Condition monitoring and fault detection

Abstract

Early attempts at monitoring the condition of railway point mechanisms employed simple thresholding techniques to detect faults, but success was limited and there were large numbers of false alarms and missed failures in the field. More recent research using data collected from line-side equipment and lab-based test rigs, though, is suggesting that it should indeed be possible to predict failures with sufficient accuracy and notice to be of genuine use to infrastructure maintainers and owners. This review into state-of-the-art predictive fault detection and diagnosis methods shows how some very different generic models have been tailored to the various types of mechanisms that are in use worldwide. In any specific case, the most appropriate combination of quantitative and qualitative techniques will be determined by the inherent failure modes of the system and the particular conditions under which it operates. Furthermore, it is vital to have a priori knowledge of the symptoms that are observable under fault conditions if diagnosis is to be reliable.