基于DDBN-Cloud的舰艇编队防空目标威胁评估方法

Threat assessment method of warships formation air defense based on DDBN-cloud model

海上编队防空作战中,针对来袭目标威胁评估过程中不确定性因素较多、观测数据易缺失以及已有评估方法难以考虑动态威胁态势的问题,提出了基于离散动态贝叶斯网络-云模型(discrete dynamic Bayesian networks cloud,DDBN-Cloud)的威胁评估方法。通过分析来袭目标特征,构建了目标威胁评估体系;为避免节点威胁属性值在小范围内连续变化所引起的重复计算,采用模糊逻辑理论将体系中的连续型变量转化为离散型变量;针对评估过程中指标数据缺失问题,采用前向信息修补算法进行信息预测修补;采用证据可信度对不确定性节点的先验概率进行赋值,使得贝叶斯网络(Bayesian network,BN)参数更贴合实际;最后,利用云模型将得到的威胁评估概率转化为确定的威胁度,实现由定性概念到定量数值的转化,进行威胁排序;仿真实验表明,该方法适用于目标数据缺失时的动态威胁评估,与静态贝叶斯网络-云模型(Bayesian networks cloud,BN-Cloud)法和相对熵排序法相比,其结果更合理,具有一定的实用价值。

In the warships formation air defense operations,in the process of threat assessment,there are many problems,such as many uncertainties,lack of observation data or errors,and the existing assessment methods are difficult to consider the dynamic threat situation.A threat assessment method based on discrete dynamic Bayesian networks Cloud(DDBN-Cloud)model is proposed.By analyzing the characteristic of incoming targets,an assessment system is constructed.The continuous variables in the network structure are transformed into discrete variables by using fuzzy logic theory,which can avoid the repeated computation caused by the continuous change of threat attribute values of nodes in a small range.Aiming at the problem of missing data in the evaluation process,a forward information repair algorithm is proposed to carry out information prediction and repair.The prior probabilities of uncertain nodes are assigned according to the reliability of evidence,which makes the parameters of BN more realistic.Finally,the Cloud model is used to transform the probability of the threat assessment to the value of the threat assessment,and then threat ranking is carried out.Simulation results show that the proposed method is suitable for dynamic threat assessment when the target data is missing.Compared with the static Bayesian networks Cloud(BN-Cloud)model method and the relative entropy ranking method,the proposed method is more reasonable and has certain practical value.