Intelligent Fault Detection and Diagnosis of Air Leakage on Train Door

Abstract

Train door is a critical subsystem in a railway system. Fault detection and diagnosis in the early stage of the train door subsystem are essential for improving pre-emptive maintenance capability and reducing train downtime. This paper presents a machine learning method for automated detection of air leakage faults occurring on train door subsystem. Fifteen features are extracted from the pressure signal in each door open and close cycles. The Multi-class Support Vector Machine with Radial Basis Function as the kernel function is used for classification. Preliminary laboratory test results suggest that the proposed method has potential to serve as an intelligent train door fault diagnosis system.