系统工程理论与实践
繫統工程理論與實踐
계통공정이론여실천
Systems Engineering—Theory & Practice
2012年
6期
1305~1313
,共null页
复杂自适应系统 动态协作 任务求解 多Agent系统 时序逻辑
複雜自適應繫統 動態協作 任務求解 多Agent繫統 時序邏輯
복잡자괄응계통 동태협작 임무구해 다Agent계통 시서라집
complex self-adaptive system; dynamic cooperation; task solving; MAS; sequential logic
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.
藉鑒組織學思想將自適應繫統中的自主運行單元抽象為Agent,把複雜自適應繫統視為多Agent組織,從時間和狀態角度對複雜動態繫統的行為進行描述,提齣瞭基于時序活動邏輯的多Agent動態協作任務求解自適應機製和構造模型;分析瞭任務求解BDI Agent的信唸、願望、意圖的產生過程和實現方法,建立瞭協商推理的語義規則和行為規則,給齣瞭協作群組的選擇算法.併從任務求解Agent的心智變化角度,描述瞭動態協作任務求解模型實現的六箇階段:任務動態分配、協作意願產生、協作群體生成、共同計劃製定、協作群體行動和結果評估.通過在MAGE等平檯上實驗和倣真測試,驗證瞭方法的可行性和有效性.
차감조직학사상장자괄응계통중적자주운행단원추상위Agent,파복잡자괄응계통시위다Agent조직,종시간화상태각도대복잡동태계통적행위진행묘술,제출료기우시서활동라집적다Agent동태협작임무구해자괄응궤제화구조모형;분석료임무구해BDI Agent적신념、원망、의도적산생과정화실현방법,건립료협상추리적어의규칙화행위규칙,급출료협작군조적선택산법.병종임무구해Agent적심지변화각도,묘술료동태협작임무구해모형실현적륙개계단:임무동태분배、협작의원산생、협작군체생성、공동계화제정、협작군체행동화결과평고.통과재MAGE등평태상실험화방진측시,험증료방법적가행성화유효성.
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.