基于GA-BP网络的LFM信号脉冲压缩  

LFM Signal Pulse Compression Based on GA-BP Network

作　　者：杨宁国[1] 任新涛[1] 

机构地区：[1]中国人民解放军63863部队

出　　处：《现代防御技术》2015年第1期76-80,172,共6页Modern Defence Technology

摘　　要：为了改善线性调频(linear frequency modulation,LFM)信号脉冲压缩输出的性能,研究了误差反向传播(back propagation,BP)神经网络在线性调频信号脉冲压缩中的应用。采用遗传算法(genetic algorithm,GA)对BP神经网络的连接权值进行训练学习,该算法可克服BP网络容易陷入局部最优的缺点。仿真结果表明,GA-BP网络具有较快的收敛速度和较好的数值稳定性,在信噪比损失小于1dB的条件下,可获得60dB左右的输出主旁瓣比。The application of error back propagation（ BP） neural network to pulse compression of LFM（ linear frequency modulation） is analyzed to improve the performance of pulse compression output of LFM. Genetic algorithm（ GA） is utilized to conduct training study on the connection weight of BP neural network and such algorithm may overcome the defect that BP network easily falls into local optimum. The simulation result demonstrates GA-BP network features a faster rate of convergence and better numerical stability,and on the condition that the SNRL is less than 1dB,the output main-to-side-lobe ratio of about 60dB may be obtained.

关 键 词：线性调频信号 脉冲压缩 遗传算法 GA-BP网络 主旁瓣比 

分 类 号：TN958[电子电信—信号与信息处理] TP183[电子电信—信息与通信工程]