A neuro-fuzzy approach to automatic diagnosis and location of stator inter-turn faults in CSI-fed PM brushless DC motors

The paper presents a neuro-fuzzy-based perspective to the automation of diagnosis and location of stator-winding interturn short circuits in CSI-fed brushless dc motors. Performance of the drive under normal and short-circuit conditions are obtained through classical lumped-parameter network models. Waveforms of the electromagnetic torque and summation of phase voltages are monitored to develop two independent diagnostic algorithms. Diagnostic indices derived from the characteristic waveforms using discrete Fourier transform (DFT) lead to identifying the number of shorted turns. Fault location is achieved through a different set of indices extracted by the short-time Fourier transform (STFT). Adaptive neuro-fuzzy inference systems (ANFIS) are trained based on simulation results to automate the diagnostic process. ANFIS testing along with the good agreement between simulated and measured waveforms show the effectiveness of the proposed techniques.