基于图模型自主优化的多无人机多目标攻击  

Autonomous optimization for multiple-target attack of multiple UCAVs based on graphical model

作　　者：郭文强[1] 高晓光[1] 任佳[1] 肖秦琨[1] 

机构地区：[1]西北工业大学电子信息学院,陕西西安710072

出　　处：《系统工程与电子技术》2010年第3期574-578,共5页Systems Engineering and Electronics

基　　金：国家自然科学基金(60774064);国家自然科学基金重大研究计划(90205019)资助课题

摘　　要：为在多无人作战飞机(unmanned combat aerial vehicle,UCAV)执行多目标攻击中适时确立决策优化的方向、改变任务优化所需的基本条件,采用图模型中的动态贝叶斯网络(dynamic Bayesian network,DBN)构建了空天威胁体感知模型,提出了基于图模型自主优化系统的分层架构和多UCAV自主协同规划方法。该方法利用数据融合形成的DBN状态转移网络及观测转移网络参数的变化表现复杂空天环境的变化,并充分利用DBN的学习和推理算法,实现了对威胁体的在线动态感知,达到了按照确定原则完成UCAV攻击目标重新分配与航迹协同等任务的目的。仿真结果表明了这种自主优化规划方法的正确性和可行性。For the sake of changing main conditions required by multi-targets attack optimization and establishing the new direction of the decision-making optimization during attacking action executing by the multiple unmanned combat aerial vehicle（UCAV）,a dynamic awareness model for threats based on a kind of graphical model,dynamic Bayesian network（DBN）,is constructed.Based on this graphical model,a novel layered conceptual framework for autonomous optimization systems and an autonomous cooperative planning method for UCAVs are advanced.With parameters derived from the fusion data and observing their change in relative DBN＇s transition networks in virtue of DBN＇s learning and inference algorithms,this method realizes the online awareness for complicated aerospace surroundings.In the light of this dynamic awareness model,attack targets re-assignment and cooperative path re-planning for UCAV are achieved in accordance with an identified guideline.Simulation results demonstrate that this autonomous planning method is proper and valid.

关 键 词：图模型 无人作战飞机 决策优化 多目标攻击 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]