基于因果发现与博弈协同的航运韧性调度方法OA
Maritime Resilient Scheduling Method Based on Causal Discovery and Game-Theoretic Collaboration
现有集中式航运调度依赖预定义事件模式,难以应对多方博弈与未知"黑天鹅"事件.针对此局限,提出一种基于分布式因果发现与博弈协同的应急决策框架(CIRCE),旨在提升航运网络在复杂扰动下的自适应韧性.首先,利用改进的在线PCMCI(Peter-Clark Momentary Condi-tional Independence)算法赋予船舶智能体从时序数据中动态推断局部风险因果图的能力;其次,设计轻量级因果信念通信协议实现分布式认知的动态对齐;最后,构建基于谢林点(Schelling Point)的非合作博弈模型,能够在无中心指令下快速形成协同减速、协同绕航、虚拟到港、接力运输等最大化集体利益的协同策略.基于真实台风"烟花"气象场与船舶自动识别系统(AIS)数据的数字孪生仿真表明,该方法通过分布式因果机制实现了对未知风险的动态感知与协同响应,克服了传统方法对中心化预置图谱的依赖,显著提升了船舶安全裕度,降低了额外运营成本,为构建高韧性智能航运系统提供了新路径.
Existing centralized maritime scheduling relies heavily on predefined event schemas,which limits its capability to address multi-party gaming behaviors and unforeseen"Black Swan"events.To address these limitations,an emergency decision-making framework based on distributed causal dis-covery and collaborative game theory(CIRCE)is proposed to improve the adaptive resilience of mari-time networks under complex disturbances.Firstly,an improved online PCMCI algorithm is utilized to empower vessel agents to dynamically infer local risk causal graphs from time-series data.Secondly,a lightweight causal belief communication protocol is designed to facilitate the dynamic alignment of dis-tributed cognitive states.Finally,a non-cooperative game model based on the Schelling Point is devel-oped,allowing rapid formation of coordinated strategies,including collaborative speed reduction,route avoidance,virtual port arrival,and relay transportation,to maximize collective benefits without central-ized directives.Digital twin simulations based on real-world Typhoon"In-Fa"meteorological fields and automatic identification system(AIS)data demonstrate that the proposed method enables dynamic perception and coordinated response to unknown risks through a distributed causal mechanism,over-coming the reliance on centralized pre-defined maps inherent in traditional methods.It significantly en-hances vessel safety margins,reduces additional operational costs,and provides a novel pathway for de-veloping highly resilient intelligent maritime systems.
张兵;余文涛;陈晓慧;刘建辉;罗善军;黄亮
信息工程大学,河南 郑州 450001信息工程大学,河南 郑州 450001信息工程大学,河南 郑州 450001信息工程大学,河南 郑州 450001河南省科学院,河南 郑州 450003武汉理工大学,湖北 武汉 430070
海洋科学
航运韧性分布式因果发现多智能体博弈应急响应时序因果数字孪生
maritime resiliencedistributed causal discoverymulti-agent game theoryemergency responsetemporal causalitydigital twin
《信息工程大学学报》 2026 (4)
399-405,421,8
国家自然科学基金(42371438)
评论