Fault Diagnosis of Rolling Bearing Based on Fast Spectral Kurtosis and PSO Optimized Multi Kernel Extreme Learning Machine

In today's industrial processes, the motor is a critical driving component. The bearing is the motor's most fundamental component and one of the most prone to failure. The bearing is responsible for a substantial portion of motor failure. The research of fault diagnostic technologies for motor bearings is of considerable practical importance in order to assure the regular operation of the motor equipment. Because of high efficiency learning speed and excellent generalization performance, the extreme learning machine (ELM) has been widely employed in the field of defect diagnostics in past few years. A multi-kernels extreme learning machine fault diagnostic model based on fast kurtosis spectrum is suggested in this study, and Particle swarm optimization (PSO) is utilized to discover the value of the model's superparameters, resulting in a comprehensive fault diagnostic model. Finally, the suggested technique is tested by rolling bearing data set of Case Western Reserve University(CWRU).