Multi-Hierarchical Dispersion Entropy: A Novel Nonlinear Measurement Approach for Fault Diagnosis of Rolling Bearing

Hierarchical dispersion entropy (HDisE)is an effective way to measure the dynamics of nonlinear signals. However, HDisE only extracts features at a single level, resulting in the loss of a large amount of deep information. In addition, HDisE is very sensitive to the length of time series. As the number of decomposition layers increases, the length of the original sequence decreases dramatically, which may lead to instability and inaccuracy of entropy estimation when dealing with short-term sequences. To solve these limitations, this paper proposes a new time series complexity measurement method, multi-hierarchical dispersion entropy (MHDisE), for feature extraction of signals. MHDisE not only enriches the hierarchical nature of information by integrating the high-frequency and low-frequency features of multiple decomposition levels, but also improves the stability and accuracy of short-time series entropy estimation. Based on the advantages of MHDisE, a fault diagnosis method for rolling bearing combining MHDisE and Support Vector Machin optimized by Spider Wasp Optimizer (SWO-SVM) was proposed. Finally, the CRWU public dataset and experimental data of rolling bearing are used for verification. The results show that the proposed method can effectively extract feature information, improve the fault identification accuracy and stability, and show significant advantages in processing short-term sequences.