Improved Deep PCA based Incipient Fault Detection with Application on CRH Traction System

The traction system reliability is of critical importance to the safety of the CRH (China Railway High-speed) trains. The data-driven based FDD (Fault Detection and Diagnosis) technologies of electric locomotive traction are urgently demanded to improve the reliability and avoid serious failures of High-Speed rail vehicles. This paper explores further results of Deep PCA (Principal Components Analysis) theory and presents a fault detection solution combing with multi modality for high-speed rail vehicle traction systems. The experiment results performed on TDCS-FIB (Traction Drive Control System-Fault Injection Benchmark) platform demonstrate that the method proposed in this paper can effectively reduce the high missing alarm rate or false alarm rate of CRH traction systems.