Diagnosis of induction motor faults using a DSP and advanced demodulation techniques

Abstract

On-line diagnosis of induction motors faults requires special, high speed hardware, such as DSP or FPGAs. Practical implementation of diagnosis algorithms in such a device must take into account the limited amount of memory available for storing sampled data, and for performing spectral analysis using the FFT. Another practical problem is the need to filter the mains component, whose leakage can hide fault harmonics, prior to compute the FFT of the current's signal. This requires the use of digital filters, that must be tuned in case of using variable speed drives that can operate the motor at different speeds. In this paper, an advanced demodulation technique that is able to eliminate the mains component with an extremely low memory requirement, based on the Teager- Kaiser energy operator, is presented. The demodulated current is footprint is down sampled, so that only 2kb of memory are needed to perform the diagnosis process. The proposed method is implemented in a DSP commercial board online diagnosis system and tested on commercial induction motors with broken bars. Finally, the results are compared with the results obtained offline using conventional Motor Current Signature Analysis method.