Roller Bearing Fault Diagnosis Method Based on Chemical Reaction Optimization and Support Vector Machine

Abstract

Support vector machine (SVM) parameter optimization has always been a demanding task in machine learning. The chemical reaction optimization (CRO) method is an established metaheuristic for the optimization problem and is adapted to optimize the SVM parameters. In this paper, a SVM parameter optimization method based on CRO (CRO-SVM) is proposed. The CRO-SVM classifier is applied to some real-world benchmark data sets, and promising results are obtained. Furthermore, the CRO-SVM is applied to diagnose the roller bearing fault by combining with the local characteristic–scale decomposition (LCD) method. The experimental results show that the combination of CRO-SVM classifiers and the LCD method obtains higher classification accuracy and lower cost time compared to the other methods.