Prediction of tram track gauge deviation using artificial neural network and support vector regression

Abstract

ABSTRACTAlong with increase in tram traffic, gradual increase in gauge deviation is inevitable. Excessive track gauge is considered as an effective parameter in aggravating the quality of ride as well as the traffic safety of tram services. To ensure a smooth ride, reduce the risk of the train derailment and lower cost of maintenance, preventive maintenance operations are necessary which demand accurate track degradation prediction modelling. In this study, the Melbourne tram network is considered as a case study. Two types of rail tracks including straight segments and curved segments are studied. In order to develop models for gauge degradation prediction, two machine learning models including artificial neural network (ANN) and support vector machine (SVM) are applied. Two indexes including the coefficient of determination (R2) and mean squared error (MSE) are used to evaluate the performance of the proposed models. According to the results, both ANN and SVM models provide acceptable and somewhat similar outcomes. However, the performance of ANN models in predicting gauge deviation of straight segments is slightly better than SVM models. On the contrary, the performance of SVM models in predicting gauge deviation of curved segments is slightly better than ANN models.