自适应的EEMD及其在滚动轴承故障诊断中的应用
Adaptive EEMD and Its Applications to Rolling Bearing Fault Diagnosis

为解决总体集成经验模态分解（EEMD）算法中存在的白噪声参数需要人为选择的问题，同时考虑到现实中难以获得大量典型故障样本的实际情况，提出了一种基于自适应总体集成经验模态分解（AEEMD）与支持向量机（SVM）的滚动轴承故障诊断法。首先在信号处理上使用AEEMD将原始振动信号分解成具有不同特征时间尺度的本征模态分量（IMF），对于不同的轴承故障来说，在不同频带内的信号能量会发生改变，因此可通过计算各个IMF的能量来实现故障特征提取；然后把IMF的能量特征值作为输入来构建支持向量机分类器模型；最后利用建立的模型对轴承的状态类型做出判别。在轴承故障实例中将AEEMD算法与EEMD算法进行对比，证明了AEEMD的分解效果更好；选用BP神经网络与SVM的诊断效果进行对比分析，表明本文中提出的方法能够更加快速准确地诊断出轴承的故障。
In order to solve the problems that white noise parameters need artificial selection in EEMD (ensemble empirical mode decomposition) and considering the situation that plenty of samples of typical faults are difficult to obtain in reality, an bearing fault diagnosis method based on adaptive EEMD (AEEMD) and SVM (support vector machine) is proposed. First of all, AEEMD is used to decompose the vibration signal into several IMFs (intrinsic mode functions) of different characteristic time scales in signal processing. To different bearing faults, the energy of signals in different frequency bands can change, so energy of IMFs can be used to realize fault feature extraction. Then the energy eigenvalues of IMFs are used as inputs to construct SVM classification model. At last, the classification model is used to distinguish the types of bearing faults. In bearing fault diagnosis experiment, AEEMD and EEMD are compared to prove that AEEMD decomposition is better, and then diagnostic performance of BP neural network and SVM is also analyzed comparatively. The result shows that the method proposed in this paper can diagnose bearing faults more quickly and accurately