An improved Cuckoo Search algorithm for LS-SVM optimization in bearing fault diagnosis based on signal decomposition and reconstruction

Abstract

To improve the accuracy and robustness of bearing fault diagnosis, this paper proposes a fault diagnosis method based on Fast Ensemble Empirical Mode Decomposition (FEEMD) and a multi-strategy improved Cuckoo Search algorithm (Improved Cuckoo Search, ICS) optimized Least Squares Support Vector Machine (LS-SVM). In the signal processing phase, FEEMD is employed to decompose the bearing vibration signal. The Pearson correlation coefficient method is then applied to remove irrelevant noise components and reconstruct the signal, thereby enhancing the effective fault features. Subsequently, multiscale permutation entropy (MPE) is extracted from the reconstructed signal to characterize its complexity and nonlinear dynamics, forming the fault feature vector. In the optimization phase, a multi-strategy improved Cuckoo Search algorithm (SC-ICS) integrating a sine–cosine update strategy and Cauchy mutation is proposed. This approach enhances the global search capability and population diversity, effectively avoiding premature convergence and local optima. The improved ICS is utilized to optimize the key parameters of the LS-SVM, thereby improving its classification performance. Finally, the proposed method is applied to bearing fault diagnosis experiments. The results demonstrate that the method achieves high diagnostic accuracy and robustness under complex operating conditions, outperforming conventional optimization algorithms and diagnostic models, which verifies its effectiveness in practical bearing fault diagnosis applications.