复杂自适应系统的MAS动态协作任务求解时序逻辑模型

The sequential logic model to solve the multi-Agent dynamic cooperative tasks of complex self-adaptive system

借鉴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;分析了任务求解BDI Agent的信念、愿望、意图的产生过程和实现方法,建立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.并从任务求解Agent的心智变化角度,描述了动态协作任务求解模型实现的六个阶段: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等平台上实验和仿真测试,验证了方法的可行性和有效性.

Solving dynamic complex problem is difficult in the theory and applied research of artificial intelligence and complex adaptive systems. Idea from histology is that the auto-run unit in self-adaptive system is abstracted to be Agent, the complex adaptive system is considered as a multi-Agent tissue. The behavior of complex dynamic systems in time and space is described. The adaptive mechanisms and structure model of solving multi-Agent dynamic cooperative tasks based on sequential active logic are proposed. The production process and realization of BDI belief, desire, intention of solving task are analyzed. The semantic rules and action rules of cooperative deduction is builded. The selection algorithm of cooperative groups is given. From the mind change of task solving Agent, the paper describes the six stages to realize the solving model of dynamic cooperative tasks. The six stages are dynamic allocation of tasks, collaboration will produce, generate collaborative groups, common planning making, collaborative groups action and evaluation of results. Experiments and simulation on MAGE and other platforms prove the feasibility and effectiveness of our proposed approaches.