基于深度学习的雷达侦察系统作战能力评估方法  

Combat Capability Evaluation Method Based on Deep Learning for Radar Reconnaissance System

作　　者：戚宗锋 王华兵 李建勋[2] QI Zong-feng;WANG Hua-bing;LI Jian-xun(The State Key Laboratory of Complex Electromagnetic Environment Effects on Electronics and Information System(CEMEE),Luoyang 471003;Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]电子信息系统复杂电磁环境效应国家重点实验室,河南洛阳471003 [2]上海交通大学,上海200240

出　　处：《指挥控制与仿真》2020年第2期59-64,共6页Command Control & Simulation

基　　金：国家自然科学基金项目(61673265);CEMEE(2019G0302);民用飞机专项科研(MJ-2017-S-38);航空科学基金(20170157001)。

摘　　要：效能评估是决策的依据,包括指标体系构建与优化、综合评估等。经典的方法是分步进行的,例如基于粗糙集的指标体系优化,综合评估方法包括层次分析法、模糊层次分析法以及神经网络等。借鉴深度学习在特征提取、非线性函数映射等方面的优点,尤其是其可以很好地模拟人类思维,将深度学习与传统的层次分析法相结合。针对指标体系的底层采样数据容易出现的数据量大、缺乏统一标度、指标间权重难以确定等问题进行了模型化处理,用深度置信网络(DBN)模型实现了底层评估数据的分类与判定。最后,结合雷达侦察系统作战能力评估对象,仿真验证了该智能评估方法的合理性和正确性。Effectiveness evaluation is the basis for decision-making, including the construction and optimization of indicator systems, and comprehensive evaluation. The classical method is carried out step by step, for example, based on the rough set index system optimization, the comprehensive evaluation methods include analytic hierarchy process, fuzzy analytic hierarchy process, and neural network. This paper draws on the advantages of deep learning in feature extraction, nonlinear function mapping, etc., especially it can simulate human thinking well, and combine deep learning with traditional analytic hierarchy process. Aiming at the problems of the underlying sampling data,such as large amount of data, lack of unified scale, and difficulty in determining the weight between indicators the indicator system is modeled, and the classification and judgment of the underlying evaluation data is realized by the deep confidence network(DBN) model. Finally, toward with the radar reconnaissance system, the simulation verifies the rationality and correctness of the intelligent assessment of combat capability.

关 键 词：效能评估 深度学习 深度置信网络 雷达侦察系统 

分 类 号：TN959.1[电子电信—信号与信息处理]