基于免疫RBF网络阵列的雷达信号分类识别方法

A Novel Recognition Method of Radar Signals Based on immune RBF Network Array

本文提出了基于免疫算法设计的径向基函数(RBF)网络阵列实现对雷达信号体制和用途的分类识别。同单个神经网络相比,这种方法克服了识别类型较多时存在的扩充、修改、维护等难点。在RBF子网络的训练中,采用结合记忆机制的递阶免疫算法确定RBF网络的隐层参数,有效折衷了系统性能和运算量之间的矛盾。实验结果表明,采用这种方法设计的雷达信号识别系统达到了很高的性能。

In this paper,radial basis function （RBF） network array designed by immune algorithm is proposed to the recognition of type and application of intercepted radar signals. Compared to single neural network, the method overcomes the difficulties of expansion, modification and maintenance when there are a large number of radar styles. In the training of RBF sub-network, hierarchical immune algorithm combined with memory mechanism is used to determine RBF hidden layer parameters ,which effectively compromise between performance and computation load. Experimental results demonstrate that the recognition system designed by the proposed method has reached very high performance.