Relevance Vector Machine을 이용한 전동기 결함진단

Abstract

Recently Condition monitoring and fault diagnosis of a motor has been received considerable attention, which can increase machinery availability and performance, reduction in consequential damage, increase in machine life, and reduction in spare parts manufacturing and less maintenance. Since engineers with expert knowledge and experience are rare, intelligent systems are necessary in the world. The aim of this study is to address the problem of detecting a motor fault and to find reliable methods for fault diagnosis This paper introduces a general Bayesian framework for obtaining sparse solutions to classify predicting, and a practical model ‘relevance vector machine’ (RVM) by Michael E. Tipping. In this study, the data is acquired from motor using accelerometer sensors and current sensors. RVM algorithm is applied for the intelligent condition classification and fault diagnosis system of the motor. Also, the feature extraction is employed using principle component analysis (PCA), independent component analysis (ICA) and selecting feature value. The results obtained shows that the RVM model has high accuracy and much less testing time. Also, results show that multi-class RVM produces promising results and has the potential for use in fault diagnosis of motor.