摘要
针对滚动轴承振动信号的非平稳特性,实际工况下难以采集大量的样本信号分析故障状态,提出基于自适应噪声的完备经验模态分解(CEEMDAN)与多尺度排列熵(MPE)相融合的故障识别方法。首先,对振动信号进行小波阈值去噪,利用CEEMDAN算法对去噪后的非平稳振动信号自适应分解,对分解后的若干个固有模式分量(IMF)计算互相关系数;然后,重构信号,计算其MPE并组成故障特征向量;最后,把特征向量输入到支持向量机(SVM)中,以识别滚动轴承的故障类型。通过对仿真信号以及实际实验数据的对比验证分析,有效证明了该方法的识别准确率比基于EMDMPE的故障识别方法提高5%,结果表明:基于CEEMDAN-MPE的滚动轴承SVM故障识别方法可以更准确地提取轴承的特征,并识别轴承的故障状态,有更强的实用性和有效性。
In view of the vibration signal' s non-stationary characteristics of the rolling bearing, and it is difficult to collect a large number of fault sample signals to analyze the problem in actual working conditions, a fault identification method based on complete ensemble empirical mode decomposition with adap- tive noise (CEEMDAN) and multi-scale permutation entropy(MPE) is proposed. First, the vibration sig- nal is denoised by wavelet threshold method, the non steady vibration signal after denoising is decomposed by CEEMDAN, the cross correlation coefficient of several IMFS after decomposition are calculated. Then the signal is reconstructed, the MPE is calculated and the fault feature vector is formed. Finally, the sup- port vector machine (SVM) is used to identify the fault status of the rolling bearing. Through the comparison and analysis of the simulated signal and the actual experimental data, it is proved that the recognition accuracy of this method is 5 % higher than that of EMD-MPE method. The results show that the fault rec- ognition of rolling bearing based on CEEMDAN-MPE-SVM can extract the characteristics of the bearings more accurately and identify the fault states of the bearing, which is more practical and effective.
作者
刘珍珍
陈志雄
陈进
LIU Zhen-zhen;CHEN Zhi-xiong;CHEN Jin(School of Aircraft Engineering,Nanchang Hangkong University,Nanchang 330063,China;School of Air Transportation,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《组合机床与自动化加工技术》
北大核心
2018年第11期57-61,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金项目(51465047)
航空科学基金项目(2014ZD56009)
江西省自然科学基金项目(2015BAB207011)