Three-stage Feature Selection and Fault Diagnosis Method of Gearbox Based on MOBPSO Algorithm Parallel Optimization

Accurately selecting the most discriminative features and optimizing recognition model parameters are crucial for fault diagnosis in complex mechanical systems. To address this problem, a feature selection and fault diagnosis framework based on the Multi-Objective Binary Particle Swarm Optimization (MOBPSO) algorithm is proposed. The raw vibration signals are first preprocessed using intrinsic time-scale decomposition (ITD) to extract key time-frequency features. The features are then sorted using a filtering algorithm and reordered by a weighting mechanism to reduce the feature dimensions. Finally, MOBPSO is applied to optimize the feature subsets and classifier parameters in parallel, thus improving the accuracy and efficiency of fault diagnosis. The experimental results show that the proposed method is significantly better than the existing methods in terms of diagnostic accuracy, demonstrating its efficiency and innovation in fault diagnosis problems.