Text mining based fault diagnosis of vehicle on-board equipment for high speed railway

Natural language in the maintenance data of high speed railway system is the big challenge for the fault diagnosis due to its unstructual feature and uncertainty semantics. In this paper, a text mining based fault diagnosis method for vehicle on-board equipment (VOBE) of high speed railway has been proposed, in which, the topic model is used to extract the fault feature from the maintenance records with the arbitrary nature. In addition, a Bayesian network (BN) is also used to adapt the uncertainty and complexity of fault diagnosis of VOBE. Furthermore, a method that fully utilizes domain expert knowledge and data is presented to derive an appropriate BN structure for VOBE. At last, the correctness and accuracy of the proposed method has been verified by the real data from Wuhan-Guangzhou high speed railway signaling systems.