Abstract: The purpose of reconnaissance system’s information processing is signal recognition，which is also a key step in whole process of radar reconnaissance information processing.In order to solve the problem of radar signal recognition，a new type of recognition model is established which combines rough set and neural network closely in this paper.The model mixes the rough set’s strong rule extraction ability and the excellent classification ability of neural network.The simulation results illustrate the model reduces subject factors in signal recognition，improves network’s structure and can manage to identify radar signal without any transcendental information objectively and effectively.

Key words: signal recognition, rough set, characteristic parameter, attribute reduction, neural network

摘要： 信号识别是侦察系统信号处理的目的，是整个雷达对抗信号处理中关键性的一个环节。为解决雷达信号识别的问题，提出将粗集和神经网络紧密结合建立新的识别模型，该模型充分融合了粗集强大的规则提取能力和神经网络优良的分类能力。实验表明，该模型减少了识别的主观因素，简化了神经网络结构，能够对雷达信号有效地识别。

关键词: 信号识别, 粗集, 特征参数, 属性约简, 神经网络