文献类型：期刊文献

英文题名：Concentric diversity entropy: A high flexible feature extraction tool for identifying fault types with different structures

作者：Wang, Xianzhi[1,2];Liu, Lishuai[3]

机构：[1]Xian Univ Posts & Telecommun, Sch Automat, Xian 710121, Peoples R China;[2]Northwestern Polytech Univ, Sch Mech Engn, Xian 710072, Peoples R China;[3]East China Univ Sci & Technol, Sch Mech & Power Engn, Shanghai 200237, Peoples R China

年份：2022

卷号：171

外文期刊名：MECHANICAL SYSTEMS AND SIGNAL PROCESSING

收录：;EI(收录号：20220711639816);WOS:【SCI-EXPANDED(收录号:WOS:000877295200001)】；

语种：英文

外文关键词：Fault diagnosis; Rotating machinery; Feature extraction; Concentric diversity entropy; Complexity

摘要：Fault diagnosis technique plays an important role in ensuring safety operations and preventing catastrophe for the rotating machinery. In the fault diagnosis framework, the entropy-based method becomes a promising tool for feature extraction. Among the entropy-based methods, the diversity entropy (DE) has arisen increasing attention due to its merits of high consistency, strong robustness, and high calculation efficiency. However, DE confronts the challenge of identifying fault types with different structures, which limits the broader application of DE in the practical engineering. Generally, when faults occur on different structures, the fault features distribute widely over the full frequency band and the main difference is the sideband around the characteristic frequency. Unfortunately, the Haar wavelet used in DE can hardly match all the oscillation pattern caused by different structures, resulting in deficient feature extraction. Hence, this paper presents a high flexible feature extraction method called concentric diversity entropy (CDE), which utilizes multiple wavelets to extract fault features over the full frequency band. Based on CDE, a diagnosis framework has been developed. At last, the simulation and experiment results show that the proposed method outperforms five entropy-based methods in identifying fault types with different structures.