Induction Motor Bearing Fault Detection using Frequency Domain PCA with SVM

Abstract

— the induction motors are the integral parts of the machine drive systems and are subject to the occurrence of various defective electrical or mechanical states. The use of diagnostic methods can precisely predict the faults in its initial stages and can reduce the downtime as well as the maintenance cost of the machine. The vibration signals analysis can be used to successfully detect the malfunctions on the rotor, stator, and other electric motor components. However, a reliable detection system must satisfy some criteria like a high true positive rate, very low false-positive rate & system must be capable of detection of faults even with natural aging signs & environmental noise. These variations are unavoidable and specialized arrangements to compensate them cam affect the flexibility of the system. In this paper, we are proposing the method for where PCA is used as a feature extractor for fixed features and variable features and SVM is used as a classifier.