Fault Detection Using On-Line Wavelet Analysis: Application Induction Motor

Abstract The wavelet transform is used the to detect on-line singular behaviours for signals having a gaussian-like distribution. Hence the information about the induction drive faults becomes available for further exploit. The wavelet coefficients bear all the information about the residuals behaviour changes induced by faults. An appropriate threshold discriminates the fault-related coefficients from the nominal functioning associated ones. The construction of the decision table takes into account the time consecution and the coincidence across decomposition levels of the threshold exceeding, therefore the alarm is persistent and gradual.