Health Assessment of Turnout System Based on Mahalanobis Distance

Abstract

This paper presents a health evaluation of S700K turnout system based on Mahalanobis distance. The main implementation steps of this method consist of obtaining a time-domain statistical index data matrix of the normal-state turnout system, performing principal component analysis on the time-domain statistical index data matrix, establishing baseline Mahalanobis distance for the turnout system normal, determining the health threshold for the safe operation, evaluating the turnout system in an unknown health state. If the Mahalanobis distance of health assessment threshold is greater than the threshold, then it is suggested that condition-based maintenance be taken. An experiment case is demonstrated to verify the efficiency of the proposed method.