Discrete Wavelet Transform Based Detection of Disturbances in Induction Motors

In this work, two discrete wavelet transform (DWT) based algorithms, which have both a detection and classification phase, are developed for diagnosing and detecting various disturbances occurring in three-phase induction motors. In the first approach, the criterion for the detection phase is the comparison of the DWT coefficients of fault currents using the selected mother wavelet `db3´ at the sixth level of resolution with a threshold determined experimentally during healthy condition of the motor. A feature vector representing the second norm of details of fault currents of six levels of resolution is used to discriminate between different faults. The second approach is based on the comparison of modulus maxima of the DWT coefficients for fault detection. The classification of faults is based on localized parameters estimation. The protection phase of both the algorithms is implemented in real-time using the ds1102 digital signal processor board. It gives a trip signal almost at the instant or within one cycle of the fault occurrence in all cases of faults