Statistical Analysis of Vibration Signals : A Predictive Maintenance Strategy for Locomotive Auxiliary Drives

The motivation for this paper is the need for improved reliability/availability of different electrical drives in a locomotive for improving the reliability and quality of service to the passengers as well as the requirement for cost-effective and more efficient condition-based maintenance management. A failure of a locomotive online not only affects the profits of the railways but also adversely affects the image of the system. To avoid failure of locomotives online, various strategies are devised but a zero failure scenario has not been achieved so far. To achieve the reliability of the highest order, various maintenance practices are being used for different components. This paper deals with the maintenance practice based on the online condition monitoring of different components so that the timely and cost-effective attention is paid to the equipment whenever it is needed. This predictive maintenance strategy can be further incorporated with some intelligent monitoring system like Artificial Neural Networks (ANN) or fuzzy logic. The embedded software would implement the ANN-based learning algorithms which model the behavior of the equipment during healthy operation and provide ‘pre-failure warning’ when there is a significant deviation from the expected performance