基于EEMD的车辆微动信号提取及分类

摘要

Aiming at difference of micro-motion echo signal of two kinds of vehicles,target vehicle is identified and classified. Ensemble empirical mode decomposition(EEMD) is employed to decompose original signal into a number of intrinsic mode function(IMF). By means of correlation analysis,it is proved that EEMD can effectively overcome the mode mixing problem caused by EMD. On this basis,four features are extracted,the nearest neighbor method is used for classification. Experimental results show that the features extracted after EEMD are effective and fairly robust against the variation of the target velocity and azimuth angle.%针对轮式和履带式车辆微动信号的差异对目标车辆进行了识别分类,利用集合经验模式分解(EEMD)将原始信号分解为多个本征模函数(IMF),通过相关性分析,验证了EEMD能够有效克服EMD所带来的模态混叠问题.在此基础上,提取了4种特征,采用最近邻方法进行分类.实验结果表明:经EEMD所提取的特征是有效的,对目标速度,以及方位角的变化具有相当的稳健性.