摘要
在传统智能诊断基础上 ,将多Agent技术应用于动态、分布、实时和不确定的复杂系统故障诊断领域 ,以求解复杂过程诊断问题 .提出了基于MAS的分布式智能故障诊断方法和过程 ;讨论了基于模式聚类的故障求解机制及对诊断问题任务的辨识、分解 ;研究了多Agent的约束和关联 ;设计了应用Agent工作状态的表达机制 ;确定了应用Agent间的工作状态影响关系及多Agent间的交互、协作和通讯 ;构建了多Agent模糊关联模型 ;给出了多Agent诊断系统局部诊断决策与全局诊断决策的集成描述结构 ;实现了一种分布式Agent诊断系统结构及其原型系统 .工程应用表明 ,该系统能快速、准确地进行故障成因分析 ,并给出有效的决策意见 ,取得了与专家相似的诊断结果 .
Based on distributed artificial intelligence,the paper put forward MAS based distributed intelligent malfunction diagnosis method and process,and discussed pattern-clustering based malfunction solution-finding mechanism and the identification and decomposing of a diagnosis task.It also studied the restriction and relationship of multi-agent in the macroscopic sense,and designed the expression mechanism for applied agent's work status.Else it ascertained the influence and the communication and the interaction among applied agents,and constructed a multi-agent fuzzy related model.The last aspect is that the paper presented an integrated descriptive structure of local diagnosis decision and global diagnosis decision,and established a distributed agent diagnosis system structure and its prototype model system too.Diagnosis system can be divided into applied agent and service agent,using distributed communication mechanism to realize intercommunication of information,realize distributed system's information exchange and diagnosis,and enhance system's efficiency,security and error-standing for problem-dealing.When the model was applied in the power enterprise's security monitor system the result was similar to expert system,and the model overcame many deficiencies of previous monitor diagnosis system,enhanced the efficiency of the enterprise security running.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2004年第F12期235-237,共3页
Acta Electronica Sinica
基金
国家自然科学基金 (No 60 3730 62 )
湖南省自科基金 (No 0 4JJ30 52 )
