基于VMD和KFCM的轴承故障诊断方法优化与研究  

Optimization and Research of a Bearing Fault Diagnosis Method Based on VMD and KFCM

作　　者：常勇[1] 包广清[1] 程思凯 陈鹏 CHANG Yong;BAO Guang-qing;CHENG Si-kai;CHEN Peng(College of Electrical and Information Engineering,Lanzhou University of Technology,Lanzhou 730050,China;Engineering Department,the University of Melbourne,Melbourne Australia 3010)

机构地区：[1]兰州理工大学电气工程与信息工程学院,兰州730050 [2]墨尔本大学工程学院,澳大利亚墨尔本3010

出　　处：《西南大学学报（自然科学版）》2020年第10期146-155,共10页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金项目(51967012);甘肃省重大科技专项基金项目(17ZD2GA010);甘肃省教育厅科研创新团队项目(2018C-09).

摘　　要：针对机械设备轴承故障振动信号具有强噪声、非线性、非平稳特性并致使故障特征信息难以提取的问题.提出了一种利用变分模态分解(Variational Mode Decomposition,VMD)和改进模糊聚类算法相结合的机械故障诊断新方法.首先,对采集的故障数据采用VMD和奇异值分解方法进行预处理,去除异常值及噪声;然后,采用核模糊C均值聚类(Kernel-based Fuzzy C-Means clustering,KFCM)算法来进行不同故障类型数据划分,通过计算分类系数、平均模糊熵和海明贴近度对其分类性能进行评估;最后,利用粒子群算法(PSO)对KFCM训练状态的参数进行优化.通过仿真分析和实验数据验证,该方法不仅表现出更优的分类性能,能精确、稳定进行故障识别,而且只需要少量样本数据进行训练,从而使诊断的工作量和诊断时间大为减少,为大型旋转机械设备在线故障诊断提供了理论依据.In order to solve the problems that the fault vibration signal of mechanical equipment bearings has the characteristics of strong noise,non-linearity and non-stationarity,and it is difficult to extract fault feature information,a new method of mechanical fault diagnosis based on variational mode decomposition(VMD)and improved fuzzy clustering algorithm is proposed in this paper.Firstly,the fault data collected are pre-processed with VMD and singular value decomposition to remove outliers and noise;and then the kernel fuzzy C-means(KFCM)clustering algorithm is used to classify different fault types,and the classification performance is evaluated by calculating classification coefficient,average fuzzy entropy and hamming closeness.Finally,the particle swarm optimization(PSO)algorithm is used to optimize the parameters of KFCM training state.A simulation analysis and experimental data verification show that this method not only shows better classification performance(it can identify faults accurately and stably),but also needs only a small amount of sample data for training,so that the workload and time of diagnosis are greatly reduced,which provides a theoretical basis for on-line fault diagnosis of large rotating machinery equipment.

关 键 词：模糊聚类 特征提取 变分模态分解 故障诊断 粒子群算法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]