基于改进层次斜率熵(IHSloE)的信号低频和高频故障特征提取方法  

Method for extracting low-frequency and high-frequency fault features of signals based on IHSloE

作　　者：许立学[1] 刘鑫[1] 关文锦 陈然 邝素琴 XU Lixue;LIU Xin;GUAN Wenjin;CHEN Ran;KUANG Suqin(Guangzhou Cigarette Factory,Guangdong China Tobacco Industry Co.,Ltd.,Guangzhou 510385,China)

机构地区：[1]广东中烟工业有限责任公司广州卷烟厂,广东广州510385

出　　处：《机电工程》2024年第7期1189-1197,1230,共10页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(52075182)。

摘　　要：采用传统的基于粗粒化处理的多尺度特征提取方法,无法提取故障信号中的高频部分的故障信息,导致其提取到的故障特征难以准确地表征滚动轴承的故障状态和动态特性,无法保证故障诊断的可靠性和准确性。针对该缺陷,提出了一种基于改进层次斜率熵(IHSloE)和随机森林(RF)的滚动轴承故障诊断方法。首先,利用改进层次化处理代替粗粒化处理,实现了信号的多尺度分析目的,基于斜率熵,提出了改进层次斜率熵的非线性动力学指标;随后,利用IHSloE方法提取了滚动轴承振动信号的故障特征,建立了表征滚动轴承故障特性的故障特征;最后,基于RF模型建立了多故障分类器,并将故障特征输入至RF分类器进行了训练和测试,以实现滚动轴承的故障识别目的;利用滚动轴承数据集进行了实验,并将其与其他的故障特征提取指标进行了对比。研究结果表明:IHSloE方法采用改进的层次化处理,能够快速有效地提取出振动信号中的高频故障特征,诊断准确率达到了99%,而特征提取时间仅为149.35 s;相较于采用粗粒化处理和层次处理的特征提取方法,其准确率至少提高了2%和1%,证明该方法适用于滚动轴承的故障诊断。The traditional multi-scale feature extraction method based on coarse-based processing could not consider the fault information of the high frequency part of the signal,which made it difficult to accurately characterize the fault state and dynamic characteristics of the rolling bearing,resulting in the reliability and accuracy of fault diagnosis.To solve this problem,a fault diagnosis method of rolling bearing based on improved hierarchical slope entropy(IHSloE)and random forest(RF)was proposed.Firstly,improved hierarchical processing was used instead of coarse-grained processing to achieve multi-scale signal analysis,and based on slope entropy,a nonlinear dynamic index called improved hierarchical slope entropy was proposed.Then,IHSloE method was used to extract the fault characteristics of the rolling bearing vibration signal,and the fault characteristics of the rolling bearing were established.Finally,a multi-fault classifier was established based on RF model,and the fault characteristics were input to the RF classifier for training and testing,so as to realize fault identification of rolling bearings.The rolling bearing data sets were used for experiments and compared with other fault feature extraction indexes.The research results show that IHSloE method can quickly and effectively extract high-frequency fault features from vibration signals by using the improved hierarchical processing,and the diagnostic accuracy reaches 99%,while the feature extraction time is only 149.35 s.Comparing with the feature extraction methods using coarse-grained processing and hierarchical processing,the accuracy is at least 2%and 1%higher,respectively,which can be applied to the fault diagnosis of rolling bearings.

关 键 词：故障信号高频部分特征 改进层次斜率熵 随机森林(RF)分类器 多尺度特征提取方法 改进层次化处理 故障诊断的可靠性 

分 类 号：TH133.3[机械工程—机械制造及自动化]