Modern methods of signal processing are increasingly gathering interest in the application of the quality test of electric motors. Not only the detection of a fault is important, but also the kind, the location, and the cause of a fault. The paper deals with applications of pattern recognition, especially the feature generation, the feature evaluation, the feature selection and the fault classification. The Blind Source Separation is a mean for the suppression of unwanted noise of the test environment. A model-based method for the determination of suitable features in the frequency domain is presented.
Print ISSN: 0171-8096
Volume: 70, 09/2003
Pages: 417