Wear diagnosis for rail profile data using a novel multidimensional scaling clustering method

Abstract

The diagnosis of railway system faults is significant for its comfort, efficiency, and safety. The rail surface wear is the key impact factor when considering the health conditions of rails. This paper accomplishes contactless rail wear diagnosis by using multidimensional scaling based on a novel informational dissimilarity measure (IDM) to cluster intact and different worn rail profile data. The IDM uses weighted‐probability distribution of dispersion patterns to extract accurate time domain features from rail profile data, and the loss of information is minimized, which can greatly improve the accuracy for wear diagnosis. All the analyzing data for real experiments are collected by a laser scanner camera on an inspection car, where heavy‐haul railway rails with different types of surface wear are inspected. Experimental results with simulated and reality‐based data show that the proposed methods can identify worn profile data and discriminate different types of worn profiles more effectively when compared with existing methods. Thus, the proposed method offers a new thinking for the diagnosis of rail surface wear for heavy‐haul railways.