Railway Transportation Efficiency Improvement: Loco Health Assessment by Time Domain Data Analysis to Support Condition Based Maintenance Implementation

Health estimate of loco mechanical transmission components usually implies the on-board installation of fit-to-purpose devices and the use of specific protocols for data acquisition. In this paper the authors present a different approach, based on the acquisition and post-processing of the stator currents of an induction traction motor, easy to be measured and already monitored by standard on-board automation systems. The use of a PWM inverter for the electrical drive of the traction unit introduces harmonics in the stator currents, and an analogous effect is due to the presence, for instance, of a bearing fault. For this reason, the authors have paid great attention to the choice of a reliable de-noising technique for the removal of the harmonic content due to the electronic conversion unit. The so cleaned stator current time series have been then processed by the Independent Component Analysis (ICA) method, in order to identify some current “features” able to predict motor condition. On the basis of a huge amount of data deriving from a laboratory test campaign, the author present and discuss in the paper the results of their study.