一种基于贝叶斯网络的威胁估计方法

为有效处理地面战场威胁估计问题,提出了一种基于贝叶斯网络的威胁估计方法。根据专家知识建立贝叶斯网络模型,通过匹配算法实现态势估计记录与贝叶斯网络中态势节点的匹配,并根据证据可信度及先验概率动态实现态势节点的概率赋值,利用Pearl消息传播算法计算威胁等级节点各取值的概率值,最终获得威胁等级决策。针对一个简化的贝叶斯网络模型,采用该方法进行了不同先验概率及态势下威胁等级的仿真评估,所得结论与人工判定结论基本吻合,表明该方法可有效地应用于地面战场威胁估计领域。

New semi-supervised learning algorithm based on support vectordate description

To process the problems of the threat assessment of ground battlefield,a Bayesian network-based threat assessment method was presented.Experts' knowledge was used to construct Bayesian network model.After matched with some situation records according to a proposed matching algorithm,the situation nodes in the bayesian network got their probability values based on evidence reliability and dynamic prior probability.The probability of each value of the threat level node was computed by propagating the probabil...