摘要
前向信息修补算法可以对离散动态贝叶斯网络的缺失数据进行预测,该算法只适用于所有观测节点是相互独立的网络,却不能处理观测节点有依赖关系网络的缺失数据。针对该算法的这一缺陷,提出了改进的前向信息修补算法,在分析离散动态贝叶斯网络的缺失数据具有二种基本形式的基础上,推导出了每种形式的相应预测公式。继而构建了用于识别威胁源离散动态贝叶斯网络的模型。仿真实验验证了改进的前向信息修补算法的有效性。
The missing data on Discrete Dynamic Bayesian Networks(DDBNs) can be predicted by the forwards information repairing algorithm,however,this algorithm can only be applied to the networks whose observed nodes are independent each other,but it can't handle the missing data on the networks whose observed nodes are dependent.To overcome its disadvantage,we proposed an improved forwards information repairing algorithm.After analyzing that the missing data on DDBNs have two basic forms,we deduced the corresponding prediction formulation for each form,next constructed menace identifying model of DDBNs.It's proved by the simulation experiment that the improved forwards information repairing algorithm is more efficient.
出处
《火力与指挥控制》
CSCD
北大核心
2012年第2期199-203,共5页
Fire Control & Command Control
基金
国家自然科学基金资助项目(60774064)
关键词
前向信息修补算法
数据修补
目标识别
离散动态贝叶斯网络
威胁源
forwards information repairing algorithm
data completion
target recognition
discrete dynamic Bayesian networks
menaces
作者简介
作者简介:陈海洋(1967-),男,陕西西安人,博士研究生,研究方向:先进控制理论及应用。