文献类型：期刊文献

英文题名：A hybrid signal enhancement framework for incipient bearing fault diagnosis in electric motors based on the BAACMD-SCSSA-MCKD

作者：Hou, Xiaopeng[1] Lyu, Bin[1] An, Yuan[1] Tang, Zhaoqun[1] Peng, Xiaorui[1] Li, Yonghong[2]

第一作者：侯晓鹏

通信作者：Hou, XP[1]

机构：[1]Chinese Acad Forestry, Res Inst Wood Ind, Beijing 100091, Peoples R China;[2]PowerDekor Flooring Jurong Co Ltd, Jurong 212411, Jiangsu, Peoples R China

年份：2025

外文期刊名：PROCEEDINGS OF THE INSTITUTION OF MECHANICAL ENGINEERS PART C-JOURNAL OF MECHANICAL ENGINEERING SCIENCE

收录：;WOS:【SCI-EXPANDED(收录号:WOS:001554370800001)】；

基金： The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by National Key R&D Program of China (No. 2023YFD2201500).

语种：英文

外文关键词：Electric motor; fault diagnosis; band-adaptive adaptive chirp mode decomposition; multi-stage cyclic kinematic deconvolution

摘要：Electric motors are primary actuators in modern manufacturing, and timely diagnosis of motor faults is crucial for ensuring production safety, improving parts quality, and controlling maintenance costs. Research on data-driven fault diagnosis approaches plays a critical role in the effective utilization of motor monitoring information. However, early failure signal of motors is faint and often coupled with a significant amount of noise, making it necessary to develop advanced data enhancement techniques. Thus, this study develops a novel diagnostic model that integrates band-adaptive adaptive chirp mode decomposition (BAACMD) and multi-stage cyclic kinematic deconvolution (MCKD) methods. Specifically, BAACMD is used to adaptively decompose fault signals, effectively removing noise and isolating the relevant fault features. Then, MCKD is employed to enhance the periodicity and cyclic behavior of fault data, making fault characteristics more distinguishable. Finally, a convolutional neural network is used to categorize enhanced fault features, improving diagnostic accuracy. We evaluated our proposed approach's performance on the motor bearing dataset and compared it to other cutting-edge techniques. Results show that our method outperforms existing deep learning methods in terms of fault diagnosis accuracy and robustness.