基于时变离散DBN的无人机任务决策模型  

Mission decision-making for unmanned aerial vehicle based on time varying discrete DBN

作　　者：戴晶帼 任佳[1] 崔亚妮[1] 李媛[1] 

机构地区：[1]海南大学信息科学技术学院,海南海口570228

出　　处：《计算机工程与设计》2016年第5期1346-1351,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61440048;61162010);国家国际科技合作专项基金项目(2015DFR10510);海南省自然科学基金项目(614227);海南大学青年基金项目(qnjj1243)

摘　　要：针对动态战场环境下无人机自主任务决策问题,提出一种基于时变离散动态贝叶斯网络的决策模型学习算法。在获得的前一时刻模型结构和参数基础上,通过粒子滤波算法动态更新模型结构和参数状态,利用实时获得的非平稳观测信息,完成当前时刻的任务决策推理。仿真结果表明,该模型学习算法能够在威胁联网的战场环境中保证无人机任务决策的有效性。Aiming at the problem of unmanned aerial vehicle（UAV）mission decision-making in the dynamic battlefield environment,a learning algorithm based on time varying discrete dynamic Bayesian network（TVDDBN）was proposed.On the basis of the model structure and parameters at the previous time epoch,the states of both structure and parameters were dynamically updated by employing particle filtering,and the inference of the current mission decision-making was achieved by using the non-stationary observation information up to current time.The simulation results show that the proposed algorithm can guarantee the effectiveness of the UAV mission decision-making under the conditions of threat netting.

关 键 词：无人机 时变离散动态贝叶斯网络 威胁联网 任务决策 粒子滤波 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]