针对星载合成孔径雷达(SAR)信号的电子对抗侦察识别问题,提出了一种基于支持向量机的星载SAR信号分类方法。该方法基于样本聚集性构建二叉树结构,解决了支持向量机的多分类问题;同时减小了二叉树结构的分类误差积累,选择高斯核函数解决样本的非线性问题,并采用遗传算法对模型参数进行优化,从而提高模型的分类性能。文中对加拿大的Radarsat-2星载SAR卫星的四种信号进行了分类仿真,并与传统的参数匹配法进行了比较,结果表明文中的方法具有较好的识别率,同时模型的泛化能力也比较强,有利于解决对星载SAR 的侦察难点问题。

To solve the problem of electronic countermeasure reconnaissance and identification for space-borne synthetic aperture radar (SAR) signals, a space-borne SAR signal classification method based on support vector machine is proposed. This method builds a binary tree (BT) structure based on the sample clustering, solves the multi-classification problem of support vector machine, and reduces the accumulation of classification errors of the binary tree structure. Gaussian kernel functions are used to solve nonlinear problems of samples, and genetic algorithm is used to optimize model parameters, thereby improving the classification performance of the model. In this paper, the four kinds signals of Radarsat-2 space-borne SAR satellite from Canada are classified and simulated, and compared with the traditional parameter matching method. The results show that the method in this paper has a good recognition rate, and the generalization ability of the model is also stronger, so it is helpful to solve the difficulty of reconnaissance on space-borne SAR.

TN957.51