A rolling bearing fault classification method based on feature optimization and transformer-SVM

Deep learning-based intelligent fault diagnosis methods have been widely applied in industrial production. However, in practical scenarios, the non-stationary characteristics and strong noise interference of bearing vibration signals significantly constrain the improvement of diagnostic accuracy. To address this issue, this paper proposes an intelligent fault diagnosis framework based on Variational Mode Decomposition (VMD) and Transformer-SVM. This method first employs the Osprey-Cauchy-Sparrow Search Algorithm (OCSSA), with minimum envelope entropy as the optimization objective, to adaptively determine and optimize VMD's mode number K and penalty factor α, thereby obtaining the optimal signal decomposition result. Multi-dimensional indicators are then extracted from the reconstructed signal to construct feature vectors. Subsequently, leveraging the transformer's powerful capability for modeling global dependencies, it mines the deep nonlinear relationships among features. Combined with the Support Vector Machine's strong generalization performance in classification tasks, it achieves accurate classification of bearing faults under complex operating conditions. Comparative experiments on two public datasets show that the proposed method outperforms several existing methods in terms of both classification accuracy and robustness, verifying its effectiveness and advancement.