RBF神经网络在二相编码雷达脉冲压缩中的应用  

Application of RBF Neural Network for Pulse Compression in Binary-Coded Radar

作　　者：朱志宇[1] 王建华[1] 

机构地区：[1]江苏科技大学电子信息学院

出　　处：《现代雷达》2005年第3期47-49,共3页Modern Radar

摘　　要：研究了径向基函数(RBF)神经网络在二相编码脉冲压缩技术中的应用,对网络的学习算法进行了改进,基于最小二乘原理调整网络隐含层至输出层的连接权值,采用自适应梯度下降方法,在线调整隐层神经元的宽度参数αi,采用 13位巴克码进行仿真,结果表明,改进的算法有较快的收敛性,可获得 45dB以上的输出主旁瓣比,提高了雷达的探测性能。The application of radial basis function(RBF) ne ural network for pulse compression in binary-coded radar is discussed in this p aper. The learning algorithm is improved by using recursive least square princip le to adjust the weight value between hidden layer and output layer, and adoptin g adaptive gradients descend method to on-line regulate the width parameter α i of neuron in hidden layer. The simulation results show that the improved algorithm has faster convergence speed, and over 45 dB main lobe to side lobera tion is acquired. As a result, detection performance of radar is enhanced.

关 键 词：二相编码 雷达 脉冲压缩技术 旁瓣 输出层 探测性能 隐含层 RBF神经网络 连接权 学习算法 

分 类 号：TN958[电子电信—信号与信息处理]