Abstract: In order to recognize rolling bearing's fault types accurately according to the optimal characteristics of fault vibration signal of rolling bearing, a rolling bearing fault diagnosis method was proposed based on orthogonal matching pursuit algorithm and the optimized wavelet kernel extreme learning machine method. The OMP algorithm was used to de-noising the vibration signal of the bearing. The wavelet packet decomposition of the signal after de-noising was used to obtain the frequency band energy, and the fault characteristics were extracted. By using an improved whale optimization algorithm based on von-neumann, the penalty factor and kernel parameter of wavelet kernel extreme learning machine were optimized to design a classifier of rolling bearing's fault types. The experimental results prove that the proposed method can accurately and effectively identify the fault type.

Key words: orthogonal matching pursuit, wavelet kernel extreme learning machine, WOA, rolling bearing