A rolling bearing fault diagnosis model based on SABO optimized VMD-WTD-SVM

Aiming at the problem that it is difficult to extract and identify the fault type when the rolling bearing fails during operation, this paper proposes a rolling bearing fault diagnosis model based on subtraction-average-based optimizer (SABO) optimized variational mode decomposition(VMD) combined with wavelet threshold denoising (WTD) to extract the fault features, and fused with support vector machine (SVM). Firstly, the selection of the key parameter combination (K, α) in the VMD is optimized by SABO using the minimum envelope entropy as the fitness function. Secondly, the fault signal is decomposed by VMD according to the obtained parameters, and the effective component among them is selected by the envelope entropy value and craggy value to be processed by WTD again, and the optimal signal component IMF is obtained after reconstruction. Finally, the nine-feature data corresponding to the best signal component IMF are calculated as the feature vector of the current signal, and input them into the SVM for training and fault identification. Compared with other methods, this model performs more outstandingly in rolling bearing fault diagnosis, and the fault identification accuracy reaches 98.666 7%, which has good practical application value.