一种复杂诊断系统的多智能体协同求解算法  

Research on Distributed Cooperation Solution Method of Complex Diagnosis System Based on MAS

作　　者：向剑伟[1] 

机构地区：[1]湖南工业大学计算机学院,株洲412008

出　　处：《武汉理工大学学报（交通科学与工程版）》2007年第4期738-741,共4页Journal of Wuhan University of Technology（Transportation Science & Engineering）

基　　金：湖南省自然科学基金项目(批准号:06JJ2033);湖南省教育厅科研基金项目(批准号:06C268)资助

摘　　要：为适应故障诊断中动态变化的环境和满足求解复杂系统全部任务的要求,研究一种基于MAS的分布式智能监控系统.以诊断决策问题的结构及建模过程的层次性为基本思想,基于诊断任务的多层次混合分解策略,提出了融合不同表达方式和推理机制的多Agent并行诊断结构,分析了管理Agent,诊断Agent和决策Agent的基本职能.探讨了实时Agent的组织和演化机制,给出了并行推理过程中的冲突消解算法,建立了具有公共属性的智能体模块化层次结构,实现了基于MAS多层分布式黑板模型的智能监控系统.工程应用效果验证了该求解的可行性和有效性.To adapt the model in reconfiguring fault diagnosing to dynamic environment and the needs of solving the tasks of complex system fully, the paper introduces multi Agent and related technology to the complicated fault diagnosis, an integrated intelligent control system is studied. Based on the thought of the structure of diagnostic decision, hierarchy in modeling, multi layer decompositionstrategy of diagnosis task, a multi Agent synchronous diagnosis federation integrated different knowledge expression modes and inference mechanisms is presented. The functions of management Agent, diagnosis Agent and decision Agent are analyzed. The organization and evolution of Agents in the system are proposed, and the corresponding conflict resolution algorithm is given, layered structure of abstract Agent with public attributes is build. System architecture is implemented based on MAS distributed layered blackboard. The real world application shows that the proposed control structure successfully solves the fault diagnosis problem of complex plants.

关 键 词：复杂故障诊断 多Agent系统(MAS) 分布式求解 协同协作 融合算法 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TP181[自动化与计算机技术—控制科学与工程]