基于免疫RBF网络的雷达信号分类识别

采用了以免疫RBF网络为子网络的神经网络阵列实现了对雷达信号体制和用途的分类识别。免疫RBF网络采用全局搜索的优化方式，克服了传统算法的固有缺陷，在收敛速度和性能上都有较大的提高；通过提取RBF网络核函数宽度的先验知识作为疫苗构成免疫算子，缩小了标准进化算法搜索空间的范围。采用神经网络阵列有效地解决了单个神经网络在雷达信号识别中训练时间长，扩充、修改、维护难等致命的弱点。仿真结果表明，在雷达参数不全的情况下，免疫RBF网络阵列对各种雷达的体制和用途都达到了较高的正确识别率。

Recognition of Radar Signal Using Immune RBF Network

The recognition of radar style is one of the important goals in radar confrontation. This paper realizes the classification of radar systems and radar applications by the intercepted radar signals using immune RBF network matrix. Through global searching methods in optimization, the immune RBF network overcomes the intrinsic shortcomings of conventional methods, and improves the convergence speed and performance of the network. Through extracting the preliminary knowledge about the width of the basis function as the vaccine to form the immune operator, the algorithm also reduces the searching space of the standard evolutionary algorithm. The neural network matrix effectively solves the serious problems to single neural network in radar signal recognition, such as the long training time and the difficulty in expansion, modification and maintenance. Computer simulation proves that the recognition rate of radar systems and radar applications by the immune RBF matrix are very high when input radar parameters are incomplete.