Fault detection techniques for effective line side asset monitoring

In this paper the results of current research into the state-of-the-art in predictive fault detection and diagnosis methods for railway line-side assets is presented. Research to date has mainly focussed on point machines, track circuits and level crossing systems. It will be argued that, through the use of examples, that the most appropriate method for robust fault detection is based around generic models that are tuned for a particular instance of an asset. Furthermore, once a fault has been detected, it is necessary to have an a priori knowledge of the symptoms that are observable under fault conditions to reliably diagnose faults.