一种非均匀信道化滤波方法  

A Method of Nonuniform Channelization Filtering

作　　者：杨君[1,2] 袁嗣杰[1] 吕镜清[2] 

机构地区：[1]装备指挥技术学院航天测控工程研究中心,北京101416 [2]信息综合控制国家重点实验室,成都610036

出　　处：《现代雷达》2010年第9期59-62,66,共5页Modern Radar

摘　　要：针对复杂电磁环境下雷达信号密集且带宽分布不均匀的问题,基于邻信道合并思想提出了一种非均匀信道化滤波方法。文中首先讨论了复调制信号的邻信道合并条件,然后利用余弦基神经网络训练出满足邻信道合并条件的低通原型滤波器,并对网络的稳定性进行了分析,给出了学习效率的取值范围,再用训练出的原型滤波器构造复调制滤波器组实现均匀信道化,最后将均匀信道化的相邻子信道输出相加即可实现通带的合并,从而构造出不同带宽和信道滤波器实现非均匀信道化滤波。通过计算机仿真,验证了该方法的有效性。An efficient nonuniform channelization method based on ACM（Adjacent Channel Merging） is proposed for extracting large amount of radar signals.The ACM condition of the complex modulated filter bank is analised at first.Then the neural net-works based on cosine basis functions are used to train the LP prototype filter in order to satisfy the ACM condition,and the stabili-ty of the networks is analised to give the value range of the learning index.Then use the LP prototype filter to form a complex mod-ulated filter bank to realize the uniform channelization,and then add the outputs of the adjacent channels to realize the nonuniform channelization.The computer simulation results are given at last to prove this method is valid.

关 键 词：信道化 调制滤波器组 神经网络 

分 类 号：TN713[电子电信—电路与系统]