Bearing fault diagnosis based on CS-VMD and SSAE

Bearing is an important part of rotating machinery. The health condition of the bearing may lead to the failure of the whole mechanical equipment. Therefore, the effective intelligent fault diagnosis method has been a research hotspot. For bearing fault diagnosis at constant speed, a variational mode decomposition optimized by the cuckoo search algorithm (CS-VMD) and Stack sparse auto encoder (SSAE) method is proposed. The one-dimensional vibration signal of the original bearing is converted into two-dimensional signal by the data expansion method, and several mode components are obtained by the decomposition of each line in 2D signal with CS-VMD. The sample entropy of each mode component is calculated, and the mode component with the largest sample entropy is selected as the feature vector. The selected feature vector is recognized by SSAE, and the fault diagnosis of constant speed bearing is realized. Finally, the experiment shows that the proposed method in this paper is effective by using the experimental data of bearing failure.