Failure diagnosis of railway assets using support vector machine and ant colony optimization method

Support Vector Machine (SVM) is an excellent technique for pattern recognition. This paper uses a multi-class SVM as a classifier to solve a multi-class classification problem for failure diagnosis. As the pre-defined parameters in the SVM influence the performance of the classification, this paper uses the heuristic Ant Colony Optimization (ACO) algorithm to find the optimal parameters. This multi-class SVM and ACO are applied to the failure diagnosis of an electric motor used in a railway system. A case study illustrates how efficient the ACO is in finding the optimal parameters. By using the optimal parameters from the ACO, the accuracy of the performed diagnosis on the electric motor is found to be highest.