Classification of induction machine faults

Abstract

This paper presents the theoretical foundation of a method for classifying current waveform events that are related to a variety of induction machine faults. The method is composed of three sequential processes: feature extraction, feature selection and classification. The proposed feature extraction tool, time-frequency ambiguity plane with kernel techniques, is new to the fault diagnosis field. The essence of the feature extraction is to project a faulty machine signal onto a low dimension time-frequency representation (TFR), which is deliberately designed for maximizing the separability between classes. A distinct TFR is designed for each class. The feature selection seeks for the optimal number of features taking correlation into account. The classifier uses a quadratic discriminant function and mahalanobis distance as distance measure. The flexibility of this method allows an accurate classification independent from the level of load. This method is validated on a 5.5-kW induction motor test bench.