Research on Rolling Bearing Fault Diagnosis Based on Variational Modal Decomposition Parameter Optimization and an Improved Support Vector Machine

Aiming at the problems of modal aliasing and poor noise resistance when processing the vibration acceleration signal of rolling bearings by empirical modal decomposition (EMD), a variational modal decomposition (VMD) method based on parameter optimization is proposed. Combined with the improved particle swarm optimization algorithm (IPSO) and improved envelope entropy, the VMD decomposition layers and penalty parameters were optimized. The components with high correlation coefficients with the original signal were screened out, and the fault characteristics were extracted by combining the sample entropy. Aiming at the low classification accuracy of the support vector machine with fixed parameters in the fault diagnosis stage and the defects of the gray wolf algorithm, such as insufficient population diversity and large influence of the initial population on the optimization effect, an improved gray wolf algorithm (IGWO) based on multistrategy improvement is proposed. The IGWO was combined with the support vector machine to obtain an improved gray wolf algorithm optimization support vector machine (IGWO-SVM). The rolling bearing fault diagnosis test bench is established to collect the vibration acceleration signals of rolling bearing under different states. The experimental results show that the fault diagnosis of rolling bearings with strong noise can be effectively realized by applying the above methods, and the average fault diagnosis accuracy rate reaches 98.875%.