Rolling bearing fault diagnosis based on modified hierarchical multi-mode slope entropy

Journal Information

J. Hefei Univ. Tech. (Nat. Sci.), Apr.2024, Vol.47 No.4: 464-471

DOI: 10.3969/j.issn.1003-5060.2024.04.006

Authors

JI Lei $ ^{1,2} $, CHEN Jian $ ^{1,2} $, LI Wei $ ^{1,2} $, CHEN Pin $ ^{1,2} $

(1. School of Mechanical Engineering, Hefei University of Technology, Hefei 230009, China; 2. Institute of Sound and Vibration Research, Hefei University of Technology, Hefei 230009, China)

Abstract and Keywords

Abstract: The feature extraction of rolling bearing vibration signal is usually difficult, and the quality of feature extraction has a great influence on the diagnosis result. In order to improve the accuracy of bearing fault diagnosis, a feature extraction method called modified hierarchical multi-mode slope entropy (MHMSE) is proposed and combined with extreme learning machine (ELM) to realize rolling bearing fault diagnosis. MHMSE employs the modified hierarchical method to extract the high and low frequency information of time series. Meanwhile, aiming at the dimension defect of slope entropy (SE), SE is extended to multi-mode slope entropy (MSE) to extract hierarchical component features. Inputting the fault feature vector extracted by MHMSE into the ELM, the bearing faults under nine working conditions can be identified. The experimental results show that the modified hierarchical method is better than the traditional hierarchical and multiscale sequence method. The diagnosis results of MHMSE are better than those of the modified hierarchical permutation entropy (MHPE), refined composite multiscale dispersion entropy (RCMDE), refined composite multiscale fuzzy entropy (RCMFE), refined composite multiscale sample entropy (RCMSE), and composite multiscale weigh ted permutation entropy(CMWPE).

Keywords: modified hierarchical multimode slope entropy (MHMSE); extreme learning machine (ELM); rolling bearing; fault diagnosis

作者信息

季磊 $ ^{1,2} $，陈剑 $ ^{1,2} $，李伟 $ ^{1,2} $，陈品 $ ^{1,2} $

（合肥工业大学机械工程学院，安徽合肥230009；2.合肥工业大学噪声振动工程研究所，安徽合肥230009）

摘要和关键词

摘要: 滚动轴承振动信号特征提取通常较为困难,特征提取的优劣对诊断结果影响较大。为提高轴承故障诊断的准确性,文章提出改进层次多模式斜率熵(modified hierarchical multi-mode slope entropy, MHMSE)的特征提取方法,并结合极限学习机(extreme learning machine, ELM)实现滚动轴承故障诊断。MHMSE利用改进层次方法提取时间序列的高低频信息,同时针对斜率熵(slope entropy, SE)的维度缺陷,将SE推广到多模式斜率熵(multi-mode slope entropy, MSE),用以提取层次分量的特征。通过将MHMSE提取的故障特征向量输入ELM,实现9种工况轴承故障识别。实验结果表明:改进层次方法要优于传统的层次、多尺度序列方法;同时MHMSE的诊断结果优于改进的层次排列熵(modified hierarchical permutation entropy, MHPE)、精细复合多尺度散布熵(refined composite multiscale dispersion entropy, RCMDE)、精细复合多尺度模糊熵(refined composite multiscale fuzzy entropy, RCMFE)、精细复合多尺度样本熵(refined composite multiscale sample entropy, RCMSE)、复合多尺度加权排列熵(composite multiscale weighted permutation entropy, CM-WPE)。

关键词: 改进层次多模式斜率熵(MHMSE)；极限学习机(ELM)；滚动轴承；故障诊断

Funding

国家自然科学基金资助项目（11604070）；安徽省科技重大专项资助项目（17030901409）

合肥工业大学学报(自然科学版)

Navigation