Rolling Bearing Fault Diagnosis Based on Parameter-Optimized ICEEMDAN and IPOA-SVM

Aiming at the problem of low accuracy of traditional rolling bearing fault diagnosis, a fault diagnosis model of parameter optimization Improved Adaptive Noise Complete Ensemble Empirical Mode Decomposition (ICEEMDAN) and Improved Pelican Optimization Algorithm (IPOA) optimized Support Vector Machine (SVM) was proposed. We employ the optimized ICEEMDAN method, along with Multiscale Fuzzy Entropy (MFE) and Linear Discriminant Analysis (LDA) for modal decomposition and feature extraction of the original bearing vibration signals. To enhance the accuracy and convergence speed of the algorithm, we introduce Tent chaotic mapping, Levy flights, and differential variation perturbation factors into the basic Pelican Optimization Algorithm. Furthermore, the improved Pelican Optimization Algorithm is utilized to optimize the penalty parameter (c) and kernel function (g) of SVM, thus enhancing the fault diagnosis accuracy of the SVM algorithm. Experimental results on a public dataset demonstrate that the proposed method achieves an overall recognition rate of 99% for fault diagnosis of rolling bearings under 10 operating conditions.