一种细节信号提取方法及其在信号处理的应用  

A subtle signal extraction method and its application in the signal processing

作　　者：李洪儒[1] 孙健 王成威 田再克[1] 郭蓓蓓 

机构地区：[1]军械工程学院四系,河北石家庄050003 [2]中国洛阳电子装备试验中心,河南洛阳471003

出　　处：《振动工程学报》2017年第3期493-501,共9页Journal of Vibration Engineering

基　　金：国家自然科学基金资助项目(51275524)

摘　　要：液压泵性能退化过程中,振动信号复杂度高、非线性强,导致特征信息微弱,为此,提出一种基于MUWDF与LCD-BSS的液压泵故障细节信号提取方法。首先,利用MUWDF对双通道振动信号进行融合,通过对不同分解层近似信号的筛选与融合,有效降低噪声及干扰成分的影响,改善重构信号中的特征信息;在此基础上,提出LCDBSS方法,利用BIC准则与互信息相结合选取独立性较高的ISC分量,通过盲源分离得到包含故障特征信息的细节信号;最后,通过对液压泵不同松靴程度下振动信号的应用分析,验证了该方法的有效性。During the degradation of the hydraulic pump, the vibration signals seem to be strongly nonlinear and the feature information is very weak. Therefore, a novel method for subtle signal extraction based upon MUWDF and LCD-BSS is proposed. First of all, dual-channel vibration signals are combined by the MUWDF algorithm. The approximate signals in various decomposition layers are selected and processed, so as to reduce the noise influences and increase feature information proportion. On this basis, the LCD-BSS algorithm is proposed for further dealing with the combined signal. The BIC and mutual information are combined to select appropriate ISC components to carry out the blind source separation for the required subtle signal containing sensitive feature information. Finally, the proposed method is verified by the pump vibration signals with various loose slipper degrees.

关 键 词：信号处理 信息融合 MUWDF LCD-BSS 细节信号 

分 类 号：TN911.7[电子电信—通信与信息系统]