Detection of torque oscillations in induction motor drives by linear discriminant analysis

Abstract

This paper analyzes the effects of the load torque oscillations in the signals of current and external stray flux and verifies that the characteristic signatures of this fault are clearly present even in an inverter-fed induction motor. In addition, a method based on the Linear Discriminant Analysis and on the calculation of the first odd harmonics of these signals is proposed for an automatic detection of this fault. The performances of the classifiers trained with these data are tested in different conditions of supply frequency and load, showing a good diagnostic ability for the current-based classifier. On the other hand, the flux-based classifiers reveal the risk of obtaining misleading results when the flux sensors are not permanently installed on the motor, due to the strong dependence of this signal on the position of the sensors used for its acquisition.