复杂环境下辐射源信号分类算法

Radiation Source Signal Classification Algorithm in Complicated Environment

为提高复杂电磁环境下辐射源信号的分类精度,提出了一种基于邻近传播(Affinity Propagation,AP)聚类和支持向量机(Support Vector Machine,SVM)的联合决策分类算法。该算法通过AP聚类剔除电磁干扰信号,同时获取空间信号初始类别数和聚类簇。为获得更好的分类效果,引入基于概率模型的SVM对聚类簇中的样本进行再分类,然后联合样本类别概率和该类别中样本个数在原聚类簇中的占比,对输入信号类别进行判别,并确定样本威胁等级。仿真结果表明,该算法在电磁信号特征交叠和复杂体制辐射源背景下可以有效地提高信号分类精度,提升武器系统在复杂环境中的适应能力。

In order to improve the classification accuracy of radiation source signal in complex electromagnetic environment,a joint decision classification algorithm based on affinity propagation(AP)clustering and support vector machine(SVM)is proposed.The algorithm removes the eletromagnetic interference signal and get the in-itial number of categories and clusters of the spatial signal.To get better classification effect,the probability model-based SVM is employed to reclassify samples in clustering clusters.Then,the probability of the sample category and the sample numbers in the proportion of the original clustering cluster are combined to verify the category of the input signals and make a confirmation of the samples threat levels.Simulation results show that the proposed algorithm can improve the signal classification accuracy in the background of electromagnetic sig-nal characteristics overlap and radiation source of complex system effectively,and it can enhance the adaptabil-ity of weapon system in complex environment.