Abstract: A detection method for rotor fault in induction motors, which is based on estimation of signal parameters via rotational invariance technique(ESPRIT) and pattern search algorithm(PSA), is presented. The performance of ESPRIT is tested with the simulated stator current signal of an induction motor with rotor fault, providing the results that ESPRIT can indeed identify accurately the frequencies of the components of interest in the simulated signal even with short-time sample. However it can not estimate the amplitudes and initial phases of those components with accuracy and stability. PSA is introduced to determine the amplitudes and initial phases of the frequency components in the simulated signal and the results are really satisfactory. Thus paves the way to detect rotor fault in induction motors by combining ESPRIT and PSA. The related experiment on an induction motor is conducted and the results demonstrate that the ESPRIT-PSA-based method to detect rotor fault in induction motors is effective even with short-time sample, as makes it a promising choice for induction motors operating with fluctuant load.

Key words: Detection, Estimation of signal parameters via rotational invariance technique, Induction motor, Pattern search algorithm, Rotor fault, event-related potential (ERP), fuzzy logic, image inference model, product image semantics, EEG