A variable-speed-condition fault diagnosis method for crankshaft bearing in the RV reducer with WSO-VMD and ResNet-SWIN

Abstract

Due to the noise interference and the weak characterization ability of the fault vibration signal of rotation vector (RV) reducer crankshaft bearing, it is difficult to obtain satisfactory results for the available fault diagnosis methods. For that, this paper proposes a variable-speed-condition fault diagnosis method with WSO-VMD and ResNet-SWIN. A signal reconstruction method with WSO-VMD was carried out, Firstly, the performance of VMD algorithm is improved by using war strategy optimization algorithm to select parameters adaptively. Then the signal is reconstructed considering the fault characteristic frequency, so as to realize the noise reduction of the signal. By using the residual network module and attention mechanism to replace the first stage of the original SWIN model, a novel ResNet-SWIN fault diagnosis model is established to enhance the feature extraction ability for the weak signal. The experiments with the constant-operating-condition and the variable-operating-condition are carried out to verify the effectiveness of the proposed method. The results show that, whether at variable-speed or constant-speed conditions, WSO algorithm has been proven to be the fastest convergence speed compared with WOA, SSA, and NGO optimization algorithms, and by the signal reconstruction with WSO-VMD, the variance evaluation indicator of the reconstructed signal has 36%, 21%, 46%, and 40%, respectively. ResNet-SWIN model has achieved the optimal diagnosis accuracy compared with SWIN, VIT, and CNN-SVM models in both variable-speed and constant-speed conditions.