一种基于多分类器融合的雷达辐射源识别方法

雷达辐射源识别传统方法只使用一个分类器进行分类,对于类别数较大、输入样本受噪声污染严重的情况很难获得好的识别效果.首先分析了雷达特定辐射源识别的可行性,分别提取信号的包络和双谱用来作为个体特征.然后将概率SVM引入雷达辐射源识别,得到分类识别的概率输出,并结合灰关联分析得到的灰关联度,采用D-S证据推理方法对两个分类器结果进行融合.仿真实验针对10种同类型雷达的脉冲数据,结果证明多分类器融合较之单个分类器能获得更高的识别率,提高了分类置信度.

Radar Emitter Recognition Based on Multiple Classifier Fusion

The conventional methods for radar emitter recognition usually use a classifier.This does not fit the circumstances,i.e.,the number of classes is greater and the signal to noise ratio is lower.Firstly,the feasibility of the specific emitter recognition is demonstrated.The envelope and the bispectra are extracted as individual features.Then,probability SVM and grey relation analysis are utilized for the radar emitter recognition,and the D-S evidence theory is used to fusion the recognition results of these t...