2023年11月第19卷第4期系统仿真技术SystemSimulationTechnologyNov.,2023Vol.19,No.4基于自适应动态贝叶斯网络的无人平台不确定推理机制研究刘朝辉1,汪晓玲2*,贺诚3,王中杰2(1.中国航空无线电电子研究所,上海200241;2.同济大学控制科学与工程系,上海201804;3.中国人民解放军93184部队,北京100076)摘要:针对不确定环境下的动态变化及信息的不完备性,提出一种基于自适应动态贝叶斯网络的不确定推理方法,而变结构使网络在不同情境下能够适应环境的变化,动态贝叶斯网络作为推理模型能够捕捉信息之间的时序关系和依赖性。因此通过自适应变结构和动态贝叶斯网络的结合,可有效处理缺失信息,提高推理的鲁棒性,实现对不确定信息的实时、准确推理,增强决策的科学性和可靠性。仿真分析结果表明,与Dempster-Shafer(DS)证据推理算法相比,基于自适应动态贝叶斯网络的不确定推理方法更适用于复杂环境下的不确定性推理。关键词:不确定性推理;自适应变结构;动态贝叶斯网络ReasoningMechanismwithUncertaintiesforUAVsBasedonAdaptiveDynamicBayesNetworksLIUZhaohui1,WANGXiaoling2*,HECheng3,WANGZhongjie2(1.ChinaNationalAeronauticalRadioElectronicResearchInstitute,Shanghai200241,China;2.DepartmentofControlScienceandEngineering,TongjiUniversity,Shanghai201804,China;3.Unit93184ofChinesePeople’sLiberationArmy,Beijing100076,China)Abstract:Aimingatthedynamicsinuncertainenvironmentandincompleteinformation,areasoningmechanismwithuncertaintiesforUAVsbasedonadaptivedynamicBayesnetworksisproposed.VariablestructuremakestheBayesnetworkcapableofadaptingtodifferentscenarios.DynamicBayesnetworkisqualifiedtocapturethesequencerelationshipanddependenceindifferentinformation.BycombingadaptivevariablestructureanddynamicBayesnetwork,theincompleteinformationisdealtwithefficiently,therobustnessofreasoningisincreased,reasoningforincompleteinformationisrealizedinrealtimeandwithaccuracy,andthescientificityandreliabilityfordecisionisenhanced.Thesimulationresultsshowthat,comparedwithDSinference,themethodproposedinthispaperismoreapplicabletouncertainreasoningundercomplicatedenvironment.Keywords:uncertainreasoning;adaptivevariablestructure;...