Current envelope analysis for defect identification and diagnosis in induction motors

Increasing demand in reliable manufacturing systems has been accelerating research in condition monitoring and defect diagnosis of vital machine components. This paper investigates defect diagnosis of induction motors, which are widely used in manufacturing systems as a source of actuation. A new approach, based on feature extraction from the envelope of the motor current instead of the motor current itself, has been investigated. This is based on the consideration that motor current envelope is effective in revealing the amplitude-modulated nature of the motor current signal. Three pattern classifiers - Naïve Bayes, k-nearest neighbor, and Support Vector Machine, have been investigated for defect classification. Experimental results have demonstrated that the new feature extraction and selection method yields a higher degree of accuracy than the traditional method for motor defect classification.