摘要
针对以往利用贝叶斯网络进行势评估时,贝叶斯网络结构和参数都是固定不变的不足,为提高态势评估准确性,提出一种变结构区间概率动态贝叶斯网络(variable structure interval probability dynamic Bayesian network,VSIP-DBN)进行态势评估的方法。给出了VSIP-DBN的定义,推导了其推理的算法,网络结构能够根据态势变化情况进行改变,并给出了结构变化的判断依据,将参数推广为区间概率的形式,同时提出了区间概率参数的学习方法。将VSIP-DBN应用于态势评估,在典型作战条件下进行仿真分析,不需要精确给出网络参数,即使出现偶然观测误差,也能够准确地评估出当前空战态势,提高了评估的灵活性。
The structure and the parameters of Bayesian network that is used for situation assessment are usually invariable in the past. In order to enhance the veracity of combat situation, a variable structure interval probability dynamic Bayesian network（VSIP-DBN） is proposed. The definition and the inference algorithm of the VSIP-DBN are given, the structure of VSIP-DBN can be varied according to the situation, and the rule of the network structure change is proposed. The parameters of the network are within the interval domain and the parameter learning method is also given. The air combat situation is assessed using VSIP-DBN. In the condition of interval probability parameter, even with incidental observation error, the simulation results show that the proposed model can accurately reflect the correct situation in the typical situations~ so the proposed model en- hance the flexibility of situation assessment.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2013年第9期1891-1897,共7页
Systems Engineering and Electronics
基金
国家自然科学基金(61174031)资助课题
关键词
态势评估
动态贝叶斯网络
区间概率
结构变化
situation assessment
dynamic Bayesian network （DBN）
interval probability
variable structure
