Abstract
Monitoring data under various conditions are obtained from railway facilities. These data are utilized to obtain the real-time conditions and to detect a malfunction using a threshold; however, they are rarely utilized to predict future conditions. Therefore, we have been trying to develop an actual prediction method to support appropriate maintenance work for an electric point machine by focusing on the locking-error position data that were stored in the monitoring system of an electric point machine. In this paper, a short-term prediction method is proposed to calculate the locking-error position for the next day. This paper also shows the validation results of the proposed method and the practical use of our method.
