DENOISING AND DETECTION OF FAULTED MOTOR SIGNAL BASED ON BEST WAVELET PACKET BASIS

Faulted motor signal acquired at actual working spot usually includes large amount of noise. Extraction of the fault characteristic information will be influenced greatly if the faulted motor signal is not effectively denoised. Based ondiscussing the fast searching algorithm of best wavelet packet basis (BWPB)adoping shannon cntropy, a new method based on BWPB is presented to denoise and detect the faulted motor signal. It is demonstrated by analysis of actual signal example that adopting BWPB produces a better signal denoising effect compared with signal denoising adopting wavelet analysis or ordinary wavelet packet analysis. Analyzing is performed on the denoised signal and the fault charateristic information is extracted. The detection results show that signal denoisingapplying BWPB method is in favor of enhancing the detection accuracy of motor faults.