Mean absolute difference approach for induction motor broken rotor bar fault detection

This paper proposes the use of mean absolute difference (MAD) technique to perform broken rotor bar fault defection in induction motors. Feature extraction is performed on several motor current frequency components according to different load conditions in order to obtain meaningful feature vectors for broken rotor bar fault detection purposes. The MAD between the predetermined reference vector and the feature vector extracted from the input current spectrum is used to determine whether the motor has fault or not. The MAD approach is simple and can be effectively used for broken rotor fault detection, provided that appropriate feature vector for fault information is used. Experimental results show that the proposed method effectively detects the broken rotor bar faults under different load conditions.