基于信息熵的复杂机械系统非线性特征提取方法研究

RESEARCH ON NON-LINEAR FEATURE EXTRACTING METHOD OF COMPLEX MECHANICAL SYSTEM BASED ON INFORMATION ENTROPY

复杂机械系统运行状态和特征参量往往具有较强的非线性和耦合性,运用信息熵定量评价方法实现系统状态参量的非线性特征提取。针对单一信号非线性特征提取问题,将小波包分解和奇异谱熵相结合,提出了多分辨率奇异谱熵模型,可实现信号不同频带内奇异特性的定量描述;针对多变量耦合复杂机械系统,建立了联合复杂度信息熵和信息传递指数模型,可分别描述多变量关联特征和信息传递特性。仿真实验验证了研究方法的有效性。

Seeing that the running states and feature parameters of complex mechanical systems always are of stronger non-linearity and coupling properties, the feature extraction of the non-linear state parameters is realized by a quantitative evaluation method based on information entropy. Aiming at non-linear feature extraction from a single signal, a multi-resolution singular spectrum entropy model based on integration of wavelet packet decomposition and singular-spectrum entropy is proposed, which can realize the quantitative description of the singular character of the signal in different frequency bands. As to multi-variables coupling complex systems, a joint complexity entropy and an information transfer index model are presented, which can separately describe the relating features and information transfer features of multi-variables. The validity of the method is shown by simulation experiments.