基于混合数据类型特征矢量的雷达辐射源识别

Radar Emitter Recognition Based on Feature Vector with Hybrid Data Types

为解决雷达辐射源识别中量测实数型多属性数据和数据库中实数型、序列型和区间型混合多属性数据无法直接进行决策识别的难题,提出了一种采用混合数据类型特征矢量的雷达辐射源识别方法。上述方法首先采用区间数点算子将数据库中各辐射源的混合型特征矢量分解成若干实数型子特征矢量,然后以灰关联系数作为待识别目标样本集与所有子特征矢量的相似测度,其次通过D-S证据理论确定待识别目标与数据库各辐射源的灰关联系数决策矩阵,最后根据与理想解的相对接近度对数据库各辐射源进行排序选优,从而实现决策识别。仿真结果表明,改进方法雷达辐射信号优化识别提供了参考。

A radar emitter recognition algorithm based on feature vector with hybrid data types is proposed. First of all, hybrid feature vector of each emitter in the database is decomposed into some subvectors with real number mul- tiple attribute data through interval number point operator. Then, gray correlation coefficient is embedded as similarity measure between the sample set of target which will be recognized and the whole subvectors. Thirdly, gray correlation coefficient decision-making matrix between target which will be recognized and emitters in the database is confirmed by D-S evidence theory. Finally, according to the relative approach degree with ideal solution, this method can sort emitters in the database and choose the optimal scheme so as to accomplish the decision-making and recognition. The feasibility and validity of the method is validated by simulations.