Application of ICCEMDAN combined with dispersion entropy in bearing fault diagnosis

In order to improve the accuracy and efficiency of bearing fault diagnosis, aiming at the problem of insufficient classification performance of the existing extreme learning machine (ELM) in fault diagnosis, an ELM optimization method based on fractional divergence entropy and improved pelican optimization algorithm (IPOA) is proposed. Firstly, the fractional-order divergence entropy is used to process the bearing vibration signal to extract the entropy value reflecting the fault characteristics. Secondly, aiming at the sensitivity of ELM in parameter selection, the improved pelican optimization algorithm is used to optimize the key parameters of ELM to improve its classification ability. The improved algorithm improves the search ability and convergence speed of the algorithm by introducing chaotic mapping, adaptive weight strategy and golden sine strategy. Through the analysis of experimental data, the results show that the improved ELM model achieves more than 90 % accuracy in fault diagnosis, which verifies the effectiveness and practicability of the proposed method in bearing fault diagnosis.