Fault diagnosis method of the on-board equipment of train control system based on rough set theory

摘要

Reliable and timely fault diagnosis system is an important guarantee of the high-speed train. Therefore, it is meaningful to do some research about it. In this paper, firstly, the Neural Network (NN) algorithm is applied to the fault diagnosis process of the on-board equipment Balise Transmission Module (BTM) unit. According to the result that the fault recognition accuracy rate is 19.72% in training stage and 19.98% in test stage, it can be concluded that NN algorithm has a poor performance in dealing with the high noise data. In order to overcome this drawback, a method which combines Rough Set Theory (RST) and NN algorithm is proposed. It has greatly improved the diagnostic precision and the fault recognition accuracy rate can be 93.32% in training stage, 97.41% in test stage. Finally, the experiment of Support Vector Machine (SVM) with excellent classification capacity is made based on the original data and the recognition accuracy is 75.3%. The distinct contrast of RSTNN and SVM further verifies the effectiveness and feasibility of Rough Set Theory in the implementation of the train control system's fault diagnosis.