Abstract
Fault diagnosis of rolling bearings, as a critical and fragile rotary support component in mechanical equipment, has been a hot issue. A rolling bearing fault diagnosis technique based on Time-Shifted Multi-scale Attentional Entropy and sparrow search optimisation algorithm with kernel-extreme learning machine (TSMATE-SSA-KELM for short) is proposed. Firstly, to address the problem of insufficient coarse-graining of multi-scale attentional entropy (MATE), a tool for measuring signal complexity - Time-Shifted Multi-scale Attentional Entropy - is proposed to construct multidimensional fault feature vectors.Secondly, the Sparrow Search Algorithm (SSA), which has strong optimization ability and fast convergence speed, is introduced to optimize the regularization coefficient C and the γ kernel function parameter γ of the Kernel-Extreme Learning Machine (KELM) in order to obtain the optimal parameter combinations, to solve the problem that the KELM parameters are difficult to be adjusted, and to construct the optimal SSA-KELM model. Finally, an example analysis is carried out using the bearing failure dataset of Jiangnan University to verify the influence of parameters and the effectiveness of the model. The results show that compared with different feature vector input and learning models (e.g., MATE, SVM, ELM, etc.), the proposed technique can achieve 99.85% accuracy, and it has potential engineering applications with fast computation speed and high diagnostic efficiency.