基于滤波机理改进的EMD及其在轴承故障诊断中的应用  

An Improved EMD Based on Filtering Mechanism and Its Application in Bearing Fault Diagnosis

作　　者：刘庆强[1] 冯凯 杨宁 张彦生 王永安[3] LIU Qing-qiang;FENG Kai;YANG Ning;ZHANG Yan-sheng;WANG Yong-an(School of Electrical Engineering and Information;School of Computer and Information Technology,Northeast Petroleum University;School of Electrical Engineering and Automation,Harbin Institute of Technology)

机构地区：[1]东北石油大学电气信息工程学院 [2]哈尔滨工业大学电气工程与自动化学院 [3]东北石油大学计算机与信息技术学院

出　　处：《化工自动化及仪表》2018年第11期879-884,共6页Control and Instruments in Chemical Industry

基　　金：国家自然科学基金项目(61673102)

摘　　要：深入阐述了经验模态分解(EMD)算法的过程,从滤波原理的角度论证了其全局低通滤波,挖掘出了同阶IMF迭代中的高通滤波特性,揭示了EMD的机理以及均值包络影响其分解效果的内在原因。为避免包络拟合带来的过冲或欠冲问题,提出了基于理性三次Hermite插值的EMD算法,并将该算法应用于轴承数据处理中。结果表明,所提算法提高了信号分解的有效性,从Hilbert边际谱线中准确提取了故障相关频率,具有较高的工程实用价值。The empirical mode decomposition （EMD） algorithm was elaborated and starting with the perspective of filtering principle, the existence of global low-pass filter was proved and the high-pass filtering characteristic of the same order IMF was excavated to reveal mechanism of EMD decomposition and intrinsic factor of the mean envelope which affecting decomposition effect of EMD algorithm. In order to eliminate the overshooting and undershooting in envelope fitting, an EMD algorithm based on rational cubic Herimte interpolation method was proposed and applied to bearing data analysis. Experimental results show that, the algorithm proposed can improve effectiveness of signal decomposition and fault-related frequencies accurately extracted from the Hilbert marginal line has higher practical utility.

关 键 词：经验模态分解 滤波原理 过冲或欠冲问题 理性三次Hermite插值法 故障频率 

分 类 号：TP306.3[自动化与计算机技术—计算机系统结构]