Study of the improvement of minimum change detection capability in the fault diagnosis of NS electric point machine using Bayesian estimation

Abstract

The purpose of this study is to identify the fault from the sensor outputs of electric point machine using Bayesian estimation. The electric point machine is equipment which changes the direction of the movement of train. When the fault caused to the machine, it may cause long time closure of traffic and it is costly in recovery work . Therefore, it needs that the automatic early detection and identification system for early recovery. Then in this research, probabilistic fault identification method using Bayesian estimation is proposed. Strain of a certain rod at the time of switching is measured and the deviation rate of the strain from normal condition is used as diagnostic parameter. In this study, diagnostic accuracy of the method is clarified and method for improvement is considered. As a result, diagnosis by the proposed method is possible. Further, by diagnosing the time zone in which the amount of deviation is large, the detection accuracy for a minimum fault is improved.