Fault diagnosis of speed-increasing turnout based on frequency domain feature extraction and GA-LSSVM

Abstract

The reliable operation of turnout is the key to ensure the safe operation of railways. In order to improve the accuracy of turnout fault diagnosis, a method combining genetic algorithm and least squares support vector machine (GA-LSSVM) is introduced into turnout fault diagnosis. Firstly, we need to analyze the turnout action power curve and extract the frequency domain features. Secondly, principal component analysis (PCA) is used to achieve feature dimensionality reduction. Thirdly, the features obtained by screening are input into GA-LSSVM to realize the fault diagnosis of speed-up turnout. Finally, an experimental study is performed by the failure sample set. The accuracy of the fault diagnosis method proposed in this paper reaches 96.875%. The research results show that the method in this paper can effectively improve the efficiency and accuracy of turnout fault diagnosis.