EEMD-PE与M-RVM相结合的轴承故障诊断方法

发布时间：2018-08-12 07:28

【摘要】：滚动轴承振动信号中包含了大量轴承运行状态信息,但是由于振动信号具有非线性和非平稳性的特点,难以充分提取振动信号中的故障特征,导致现有基于模式识别的轴承故障诊断方法的故障识别准确率较低.为了提高滚动轴承故障识别的准确率,提出了一种基于集合经验模态分解-排列熵(EEMD-PE)特征提取与多分类相关向量机(M-RVM)相结合的轴承故障诊断方法.首先,该方法利用EEMD对非线性和非平稳信号的自适应分解能力,将轴承故障信号分解为一组包含故障特征的本征模态函数(IMFs).然后,利用排列熵提取由EEMD分解得到的IMFs中的故障特征,并组成特征向量.最后,采用EEMD-PE对不同故障状态下的训练样本集进行特征提取,组成特征向量集对M-RVM分类器进行建模,以概率输出的形式实现对滚动轴承的故障诊断.实验结果表明:EEMD-PE特征提取方法能够对滚动轴承振动信号的故障特征进行有效提取,M-RVM能够对故障滚动轴承振动信号包含的故障特征进行识别.与现有轴承故障诊断方法相比较,所提出的方法能够提高故障识别准确率,达到99.58%.
[Abstract]:The rolling bearing vibration signal contains a lot of bearing running state information, but because the vibration signal has the characteristics of nonlinearity and non-stationarity, it is difficult to fully extract the fault characteristics of the vibration signal. The fault recognition accuracy of the existing bearing fault diagnosis methods based on pattern recognition is low. In order to improve the accuracy of rolling bearing fault identification, a bearing fault diagnosis method based on set empirical mode decomposition, permutation entropy (EEMD-PE) feature extraction and multi-classification correlation vector machine (M-RVM) is proposed. Firstly, using the adaptive decomposition ability of EEMD for nonlinear and non-stationary signals, the bearing fault signals are decomposed into a set of intrinsic mode functions (IMFs).) with fault characteristics. Then, the permutation entropy is used to extract the fault features from the IMFs decomposed by EEMD, and the feature vector is formed. Finally, EEMD-PE is used to extract the feature of the training sample set in different fault states, and the feature vector set is formed to model the M-RVM classifier, and the fault diagnosis of rolling bearing is realized in the form of probability output. The experimental results show that the fault features of rolling bearing vibration signals can be effectively extracted by the method of extracting the fault features of the fault bearing by using the component EEMD-PE feature extraction method, and the fault features contained in the vibration signals of the rolling bearings can be identified by M-RVM. Compared with the existing methods of bearing fault diagnosis, the proposed method can improve the accuracy of fault identification and reach 99.58.
【作者单位】：哈尔滨工业大学电气工程及自动化学院;
【分类号】：TH133.33

【相似文献】