Prototype Selection Method for Vehicle Condition Monitoring using Machine Learning

Abstract

The authors are developing a condition monitoring system using vibration analysis and machine learning for the purpose of monitoring the condition of railway vehicle equipment. In railway vehicles, vibrations change due to long-term state change, so long-term data should be used for learning. In this case, it is not practical to use all data, so it is necessary to use only some part of the data, which is called prototype data. Therefore, a prototype selection method based on the neighborhood method is proposed in this paper. As a result of applying the proposed method to the vibration data during the abnormal simulation test, the expected effect was confirmed.