混合贝叶斯网络在电磁态势估计中的应用  

Electronic Warfare Situation Assessment Using Mixed Bayesian Networks

作　　者：张波[1] 

机构地区：[1]解放军电子工程学院

出　　处：《微计算机信息》2009年第22期230-232,共3页Control & Automation

摘　　要：根据电磁态势估计原理,建立含连续值结点和离散值结点的混合贝叶斯网络模型,对一级融合的输出数据进行二级处理,估计干扰前后单架飞机对单个保护目标的威胁的变化情况。首先确定混合贝叶斯网络的结构,然后通过生成服从Dirichlet分布的随机数获得离散值输入结点的先验概率,根据卡尔曼滤波的结果(状态估计向量、估计协方差)确定连续值输入结点的先验概率;在连续值结点均服从线性高斯分布的假设下,通过对非线性、不连续函数进行一阶泰勒级数展开确定中间结点和输出结点的条件概率分布。仿真实验表明,混合贝叶斯网络是解决电磁态势估计问题的一种有效方法,且威胁估计的精度高。In this paper we present a mixed Bayes net based electronic warfare situation assessment algorithm from a tracking and data fusion module. The threat posed by a given intruder on a specified asset is evaluated under no-jamming condition and jamming condition. First, the inference diagram is estabilished, then the input priors of the discrete variable and continuous variable are generated from a Dirichlet distribution and from the trajectory estimate, respectively. Further, we derive approximate conditional probability distribution （CPD） for the intermediate node and output node based on the linear Gaussian assumption. The experiments indicate that the mixed Bayes net is practicable and effective.

关 键 词：混合贝叶斯网络 电磁态势估计 融合 

分 类 号：TP393.092[自动化与计算机技术—计算机应用技术] TP274.2[自动化与计算机技术—计算机科学与技术]