A Hybrid SVM-WOA Approach for Intelligent Fault Diagnosis Applications

Abstract

Swarm Intelligence is an important sub-field of Artificial Intelligence as its effective results for optimization problems. Intelligent optimization approaches employed by Swarm Intelligence techniques has attracted researchers’ interest widely and there has been also a remarkable interest in developing new techniques (algorithms). One of the most remarkable applications of such optimization techniques has become using them to train other Artificial Intelligence based techniques and form a hybrid model for better solutions. Moving from that, this study aims to use Whale Optimization Algorithm (WOA) for training a non-linear Support Vector Machines and develop an effective hybrid system of SVM-WOA for intelligent fault diagnosis. As a recent technique, WOA inspires from social behaviors shown by humpback whales and ensures a general optimization framework. Briefly, the optimization considered in this study was done to adjust sigma parameter of the Gaussian kernel function and the most optimum structure for SVM, which can provide better diagnosis results were obtained in this way. In the study, the SVM-WOA system was applied in some known fault diagnosis applications and positive findings were obtained with them. The paper briefly focuses on the technical background of the SVM-WOA system and considers its solutions for fault diagnosis.