摘要
介绍了将神经网络方法应用于通信信号的分选和识别的初步研究结果;选择了二值自适应共振(ART1)神经网络完成对输入信号的分类,确定输入信号类型是否已被网络存储,发现新出现的信号并标记;再采用多层前馈误差反向传播(BP)神经网络完成每一标记信号的识别,即识别该信号类型。比较了神经网络分类识别器和树形分类器的性能,并给出了计算机模拟结果。结果表明,基于神经网络的分类识别器的性能远优于传统技术分类器。
Military communication signal calssification and identification is one of important problem in communication countermeasures.In this paper.ART1 neural networks is used for classifying the input signal,signing the new signal and BP neural neword for identifying the signal types.This paper compares the performance of a neural net classifier with a template-based technipue tree classifier,and discusses its advantages and disadvantages.It is found that the nerual ner classfier has much more higher performance than the conventional classfier.
出处
《电子科技大学学报》
EI
CAS
CSCD
北大核心
1995年第4期350-354,共5页
Journal of University of Electronic Science and Technology of China
基金
国家"八五"预研基金
关键词
通信信号
特征提取
神经网络
分类
识别
communication signal
feature extracting
feature extracting
neural network
classification
identification