Experimental Diagnosis of Broken Rotor Bar Faults in Induction Motors at Low Slip via Hilbert Envelope and Optimized Subtractive Clustering Adaptive Neuro-Fuzzy Inference System

Knowledge of the distinctive frequencies and amplitudes broken rotor bar (BRB) faults in induction motor (IM) is essential for most fault diagnosis methods. Fast Fourier transform (FFT) widely applied to diagnose within BRBs. However, this method does not provide satisfactory results if it directly stator current signal at low slip because a high-resolution spectrum required separate different components frequency. To address problem, paper proposes an efficient based on Hilbert fast (HFFT) approach, which used extract envelope from using (HT) slip. Then, analyzed obtain amplitude frequency particular harmonic. These data were recently collected selected as BRB features employed adaptive neuro-fuzzy inference system (ANFIS) inputs autodiagnosis classification. identify defect by determining number bars rotor, two ANFIS models are proposed: grid partitioning (ANFIS-GP) ANFIS-subtractive clustering (ANFIS-SC). validate effectiveness proposed method, three motors during experiments under various loads; first was with one bar, second adjacent bars, third healthy motor. The obtained confirmed robustness combination HFFT-ANFIS-SC quantify load conditions (under high slip) precisely minimal errors (this had MSE 10-14 10-7 RMSE) compared HFFT-ANFIS-GP.