一种新型动态贝叶斯网络及其在威胁评估中的应用  

A New Kind Dynamic Bayesian Network and Application to Threat1 Assessment

作　　者：刘振[1] 彭军[1] 胡云安[1] 

机构地区：[1]海军航空工程学院,山东烟台264001

出　　处：《火力与指挥控制》2014年第2期16-20,共5页Fire Control & Command Control

基　　金：国家自然科学基金资助项目(61174031;60704090)

摘　　要：针对利用基本贝叶斯网络进行威胁评估时的不足,提出利用变结构区间概率动态贝叶斯网络(VSIP-DBN)进行威胁评估。对所提出的VSIP-DBN给出了其推理算法。在充分考虑目标的空战态势及空战能力下,建立了威胁评估的VSIP-DBN模型。将观测到的数据利用VSIP-DBN推理算法进行推理,推理结果就是目标的威胁等级。以往的威胁评估需要精确获得目标的模型参数和观测数据,通过仿真对比分析,利用VSIP-DBN建立的威胁评估模型不需要精确给出模型参数,即使出现观测数据的误差,仍能准确评估出目标的威胁程度。Aiming at the shortcoming of traditional Bayesian network that used to threat assessment, a new threat assessment method based on varied- structure interval probability DBN （VSIP-DBN） is proposed. The basic idea and definition of VSIP-DBN is proposed, the inference algorithm is also given. The threat assessment VSIP-DBN model can be constructed in view of the air combat situation and air combat capability. The observation data can be used to infer using the VSIP- DBN inference algorithms, and the inference result is just the threat level. The former methods of threat assessment must get the accurate observation data of the target and the accurate model parameter, the simulation results show that the threat assessment model based on VSIP-DBN can get the threat level even with inaccurate model parameter and incidental observation error.

关 键 词：威胁评估 区间概率 动态贝叶斯网络 

分 类 号：TP15[自动化与计算机技术—控制理论与控制工程] E955[自动化与计算机技术—控制科学与工程]