Chaos-enhanced Stochastic Fractal Search algorithm for Global Optimization with Application to Fault Diagnosis

Abstract

Support vector machine (SVM) has been known as one-state-of-the-art pattern recognition method. However, the SVM performance is particularly influenced byits parameter selection. This paper presents the parameter optimization of an SVM classifier using chaos-enhanced stochastic fractal search (SFS) algorithm to classify conditions of a ball bearing. The vibration data for normal and damaged conditions of the ball bearing system obtained from the Case Western Reserve University Bearing Data Centre. Features based on time and frequency domains were generated to characterize the ball bearing conditions. The performance of chaos-enhanced SFS algorithms in comparison to their predecessor algorithm is evaluated. In conclusion, the injection of chaotic maps into SFS algorithm improved its convergence speed and searching accuracy based on the statistical results of CEC 2015 benchmark test suites and their application to ball bearing fault diagnosis.