摘要
本文在研究带偏差单元内部回归神经网络(InternallyRecurrentNet,IRN)算法的基础上,分析了雷达特征参数和雷达威胁类型的内在联系;利用统计分析和特征建模的方法获取先验知识、设计分类器并求取模糊隶属函数,结合IRN自学习特性和模糊隶属函数的分类功能识别出目标雷达辐射源可能的威胁类型,给出识别可信度。
A radar emitter recognition algorithm based on Internally Recurrent Net （IRN）is presented . The latency incidence relation of radar parameters and radar threat type is studied. Statistic analysis and feature modeling techniques are used to conclude apriority knowledge and induce classifiers and fuzzy subjection functions. Considering radars＇multi - feature characteristics, the algorithm integrates the classifying function of fuzzy subjection function and fast convergence characteristic of IRN to output expectation threat type and its reliability.
出处
《电讯技术》
2005年第4期188-192,共5页
Telecommunication Engineering
关键词
雷达电子对抗
神经网络
辐射源识别
特征提取
特征分析与建模
知识库
Radar ECM
Neural network
Emitter recognition
Feature extraction
Feature analysis ＆ modeling
Knowledge database
