Rule-based classification approach for railway wagon health monitoring

Modern machine learning techniques have encouraged interest in the development of vehicle health monitoring systems that ensure secure and reliable operations of rail vehicles. In an earlier study, an energy-efficient data acquisition method was investigated to develop a monitoring system for railway applications using modern machine learning techniques, more specific classification algorithms. A suitable classifier was proposed for railway monitoring based on relative weighted performance metrics. To improve the performance of the existing approach, a rule-based learning method using statistical analysis has been proposed in this paper to select a unique classifier for the same application. This selected algorithm works more efficiently and improves the overall performance of the railway monitoring systems. This study has been conducted using six classifiers, namely REPTree, J48, Decision Stump, IBK, PART and OneR, with twenty-five datasets. The Waikato Environment for Knowledge Analysis (WEKA) learning tool has been used in this study to develop the prediction models.