Predictive Maintenance of Railways Turnout System Based on Vibration Data Fadhil Hidayat, Soetani, Enrico Rustam
School of Electrical Engineering and Informatics Bandung Institute of Technology
Smart City Community and Innovation Center Bandung Institute of Technology
Abstract
A turnout system is an instrument of railroad signaling equipment that directs trains to a specific lane. The turnout system is made up of several parts, including a point machine, a wessel tongue, gears, a wessel handlebar, and electrical components. Because it acts as a motor to move the wessel tongue, the point machine is the most critical component of the turnout system. Failure of a point machine to function can result in a tragic train disaster, hence its condition must continuously be monitored. An Early point machine damage detection, also known as predictive maintenance, must be performed before failure occurs. In this study, the point machine has a gyrometer sensor attached to the outside- the sensor will determine whether the vibrations in the point machine are normal or abnormal. Data is collected by observing the normal state of the point machine and then recreating the abnormal condition of the point machine with a simulation machine. This data was analyzed using Machine Learning, with 80% training data and 20% testing data. The accuracy of Machine Learning techniques such as linear regression, logistic regression, support vector regression, decision trees, random forests, and neural networks is assessed.
Keywords: Predictive maintenance, turnout system, point machine, Machine Learning
