Hierarchical Improved Diversity Entropy for Intelligent Fault Diagnosis of Rolling Bearings

Adequate extraction of feature information from nonlinear dynamic signals is the difficult and the most central part of realizing intelligent fault diagnosis of rolling bearings. The existing diversity entropy (DE) has excellent characteristics, but it only considers the relationship between the angles of neighboring orbits in phase space, and there are still inherent defects in feature extraction, which affects the diagnostic accuracy. In order to solve this problem, this paper firstly proposes the improved diversity entropy (IDE), and further proposes the hierarchical improved diversity entropy (HIDE) by combining the idea of hierarchical decomposition. Then, experiments with different types and degrees of rolling bearing failures are designed to compare and validate with other three advanced entropy-based methods. The experimental results show that compared with hierarchical diversity entropy (HDE), hierarchical weighted permutation entropy (HWPE), and hierarchical dispersion entropy (HDispE), the HIDE proposed in this paper has more prominent feature extraction capability and fault diagnosis accuracy. It also provides an efficient tool for intelligent fault diagnosis of rolling bearings.