Vibration-Based Fault Diagnosis for Railway Point Machines Using Multi-Domain Features, Ensemble Feature Selection and SVM
Yuan Cao, Yongkui Sun, Peng Li, Shuai Su
Abstract
As one of the important devices in railway signaling system, railway point machines have a great influence on train operation safety. To realize the fault diagnosis of railway point machines, this article presents a vibration signal-based diagnosis method considering its advantages of easy-to-collect and anti-interference ability. First, the vibration signals are preprocessed using Variational Mode Decomposition (VMD) for stationary preprocessing. Then a multi-domain feature extraction method is developed, which is verified as a more effective feature extraction tool than single-domain feature extraction methods. An ensemble feature selection strategy is proposed for feature selection, superior to single feature selection method. Finally, Support Vector Machine (SVM) is used for diagnosis and analysis. The diagnosis accuracy using the presented method reaches 100%, and its superiority is verified by many experiment comparisons.