An algorithm for detecting faults in railway point mechanisms

Abstract

This chapter implements a system capable of detecting faults in point mechanisms of railway infrastructure. This is achieved by developing an algorithm that takes advantage of three empirical criteria simultaneously capable of detecting faults from records of measurements of force against time. The system is dynamic in several respects: the base reference data is computed using all the curves free from faults as they are encountered in the experimental data; the algorithm that uses the three criteria simultaneously may be applied in online situations as each new data point becomes available; and recursive algorithms are applied to filter noise from the raw data in an automatic way. Encouraging results are found in practice when the system is applied to a number of experiments carried out by an industrial sponsor. A case study is presented that shows how the algorithm works with the raw data, as well as with filtered data using a particular state space (SS) system, estimated by maximum likelihood with the help of the well-known recursive algorithms Kalman Filter and Fixed Interval Smoothing.