Application of modified LMS algorithm to induction motor bearing fault diagnosis
Xiangying Duan · Dianli zidonghua shebei · 2008
It is difficult to realize reliable detection in induction motor bearing fault diagnosis by MCSA(Motor stator Current Signature Analysis) when the voltage harmonics of power supply are taken into account and the three -phase voltages are unbalanced. The conventional spectrum analytical method of stator current is studied and the weakness of conventional LMS (Least -Mean -Square) algorithm in real -time signal processing is discussed,based on which and by the perfect combination of the wavelet transform,continuous subdivision Fourier transform and modified LMS self -adaptive filter algorithm,a scheme to reliably detect induction motor bearing fault is proposed. It can correctly identify the characteristic frequency of bearing fault and greatly improve the diagnosis effectiveness. Experimental results show its feasibility.