Application of MCKD and improved LSSVM in fault diagnosis of rolling bearing

Abstract

Aiming at the problem that the fault is hard to identify during the operation of the rolling bearing, a fault diagnosis method of maximum correlation kurtosis deconvolution and improved least squares support vector machine is proposed. First of all, the method uses the maximum correlation kurtosis deconvolution to extract bearing vibration signals under different operating conditions. Then the least square support vector machine is used to supervise the extracted vibration signals. At the same time, improved cuckoo search algorithm is used to solve the problem that the kernel parameters and penalty factors of LSSVM fall into the local optimum and the convergence accuracy is poor in the optimization process, and improve the recognition rate of fault diagnosis. Bearing data were measured in different running states to verify the effectiveness of the method. The experimental results show that improved algorithm can effectively identify all types of rolling bearing status with high accuracy. It was a reliable method of bearing fault diagnosis.

Keywords

• fault diagnosis
• rolling bearing
• maximum correlation deconvolution
• least squares support vector machine
• cuckoo search algorithm