Hierarchical diversity entropy for the early fault diagnosis of rolling bearing

Intelligent fault diagnosis provides great convenience for the prognostic and health management of rotating machinery. Recently, multiscale diversity entropy has been proven to be a promising feature extraction tool intelligent diagnosis. Compared with existing methods, advantages high consistency, strong robustness, calculation efficiency. However, encounters challenge extract features from early signals weak symptoms noise. This can attributed that only concerns information embedded in low frequency, which ignores hidden frequency. To address this defect, hierarchical (HDE) is proposed, synchronously both frequencies. Based on HDE random forest, novel frame proposed. The effectiveness proposed method evaluated through simulated experimental bearing signals. results show best ability compared sample entropy, permutation fuzzy entropy.