复杂系统的MAS动态协作求解时序逻辑模型

摘要

借助于组织学思想,将自适应系统中的自主运行单元抽象为Agent,把复杂自适应系统视为多Agent组织,从时间和状态角度,对复杂动态系统的行为进行描述.提出了基于时序活动逻辑的多Agent动态协作任务求解自适应机制和构造模型；详细分析了任务求解BDIAgent的信念、愿望、意图的产生过程和实现方法；深入讨论了协商推理的语义规则和行为规则；给出了协作群组的选择算法,包括从群组的建立、选择任务Agent、分解和分配子任务；从任务求解Agent的心智变化角度,详细描述了动态协作任务求解模型实现的6个阶段:任务动态分配、协作意愿产生、协作群体生成、共同计划制定、协作群体行动和结果评估.通过在MAGE等平台上的实验和仿真测试,验证了方法的可行性和有效性.%From the ideas of tissue,the running unit of the self-adaptive system was abstracted as Agent.The complex adaptive system was regarded as a multi-Agent organization. From the perspective of time and status the behavior of the complex dynamic systems was described. Temporal logic multi-Agent based dynamic adaptive mechanisms and the task solving cooperation structure model were proposed. Detailed a-nalysis was made of the task to solve BDI Agenfs belief, desire, intention and realization of the production process. The negotiation of rules and semantic inference rules of conduct was fully discussed. The collaborative group selection algorithm was presented, including the establishment from the group. Selection of the Task Agent, decomposition and allocation of subtasks. From the changing angle of the Agenfs mind to solve the task, a detailed description was given of the task to solve the model to achieve dynamic collaboration of six stages: dynamic allocation of tasks, collaboration production, collaborative groups generation, common planning, collaborative groups, action and evaluation of results. Experiments and simulation tests on MAGE and other platforms verify the feasibility and effectiveness of the proposed approaches.