基于IMF能量矩特征提取方法

针对辐射源中的信号识别问题中特征提取困难,提出了一种基于总体平均经验模式分解的IMF能量矩特征提取方法,并且与K最近邻分类相结合用于信号识别。首先利用总体平均经验模态分解方法把信号分解成为若干个IMF,再将重要的IMF分量关于时间轴的积分,得到IMF能量矩的特征向量,最后借助K最近邻的分类(KNN)能力对特征向量进行分类,文中对十类FSK仿真信号分类表明,该方法能够有效、准确地识别信号。

Feature Extraction Based on IMF Energy Moment

Feature extraction of radiation sources is difficult in the signal recognition process;the paper presents the tMF energy moment feature extraction method based on EEMD and K nearest neighbor classification.First,the signal is decomposed into a number of the IMF,then important component of the IMF was accumulated,and finally was used to classification with the K nearest neighbor classification （KNN）.The validity of the method is verified by the experiment data.