基于变分模态分解和精细复合多尺度均值散布熵的轴承故障诊断  

Bearing fault diagnosis based on VMD and RCMMDE

作　　者：张婕 张梅[1] 陈万利 ZHANG Jie;ZHANG Mei;CHEN Wan-li(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan 232001,China)

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001

出　　处：《机电工程》2023年第5期682-690,共9页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(51874010);安徽高校自然科学研究基金资助项目(KJ2020A0309)。

摘　　要：为充分提取非线性、非平稳的轴承故障信号特征信息,进而提高轴承故障诊断精度,提出了一种基于变分模态分解(VMD)和精细复合多尺度均值散布熵(RCMMDE)的轴承故障诊断方法(算法)。首先,使用VMD将轴承故障振动信号分解为了多个模态分量,通过评估原信号与模态分量信号的互相关程度,筛选了其有效模态,并对其进行了信号重构,实现了故障信号的降噪处理目的;然后,使用精细复合均值化代替了传统粗粒化方法,利用RCMMDE方法提取了重构信号的多尺度熵值,构成了特征样本集;最后,通过鲸鱼算法(WOA)对支持向量机(SVM)进行了超参数寻优,构建了最优的故障检测模型,并将特征样本集输入到WOA-SVM模型中进行了轴承故障诊断,并通过实验评估验证了模型的有效性。研究结果表明:该模型准确率达到99.67%,精确率、召回率等各项性能指标均在99%以上,并具有很强的鲁棒性。In order to fully extract the feature information of nonlinear non-stationary bearing fault signals and improve the accuracy of bearing fault diagnosis,a bearing fault diagnosis algorithm based on variational mode decomposition(VMD)and refined composite multiscale mean distribution entropy(RCMMDE)was proposed.Firstly,the algorithm used VMD to decompose the vibration signal into multiple modal components.The effective modes were screened by evaluating the cross correlation between the original signal and the modal component signal,and the signal was reconstructed to achieve signal noise reduction.Then,the traditional coarse graining method was replaced by the fine compound mean.The RCMMDE method was used for extracting the multiscale entropy value of the reconstructed signal,and constituting the feature sample set.Finally,the whale optimization algorithm(WOA)was used for optimizing the hyperparameters of support vector machine(SVM),and the optimal fault detection model was obtained.The feature sample set was inputted into the WOA-SVM model for bearing fault diagnosis.The validity of the model was evaluated by experiments.The research results show that the accuracy of the proposed model can reach 99.67%,precision rate,recall and other performance indicators are respectively above 99%,and the model has strong robustness.

关 键 词：轴承故障诊断 变分模态分解 精细复合多尺度均值散布熵 鲸鱼算法 支持向量机 超参数寻优 

分 类 号：TH133.3[机械工程—机械制造及自动化]