베이지안 알고리즘을 활용한 열차의 고장 예측 및 FRACAS 연동을 통한 예방 정비 방법

Abstract

In order that complex system, such as train, is important to secure availability on service, reliability and maintainability on cost and safety, the efficient management method on those measures is necessary. According to the necessity, maintenance system which anticipate life-time of a train from failure data has been developed domestically. However, the method has limitation on aspect of preventive maintenance. For preventive maintenance, real-time analysis and prediction of train data are required, but train data is too much to handle in real time due to complexity of modern train system, so large amount of resources and considerable efforts are required to analyze. In order to overcome such difficulties, efficient method on handling the real time data is required. In this paper, Bayesian statistics is introduced to handle the huge amount of real time data efficiently. Bayesian statistics is utilized to generate a pattern of status information of a train. The pattern, then, is used to find out anomalous cases as failure symptom. In addition, the anomalous cases are transferred to FRACAS, FRACAS is possible to make maintenance instruction for those cases as means of effective preventive maintenance.