华南理工大学学报(自然科学版) ›› 2007, Vol. 35 ›› Issue (9): 107-112.

• 计算机科学与技术 • 上一篇    下一篇

基于多Agent 的危险源分布式监控系统

周剑峰1 陈国华1 万木生2 陈清光2   

  1. 1.华南理工大学 工业装备与控制工程学院,广东 广州 510640; 2. 广东省安全科学技术研究所,广东 广州 510620
  • 收稿日期:2006-11-06 出版日期:2007-09-25 发布日期:2007-09-25
  • 通信作者: 周剑峰(1970-) ,男,博士生,主要从事过程装备安全可靠性及风险评价技术研究. E-mail:zhou_jianfeng@sohu. com
  • 作者简介:周剑峰(1970-) ,男,博士生,主要从事过程装备安全可靠性及风险评价技术研究.
  • 基金资助:

    广东省科技计划项目(2005B32401008 )

Multi-Agent-Based Distributed Monitoring and Controlling System for Hazard Installations

Zhou lian-fengChen Guo-hua1  Wan Mu-sheng2  Chen Qing-guang2   

  1. 1. School of Industrial Control and Equipment Engineering , South China Univ. of Tech. , Guangzhou 510640 , Guangdong , China;2. Guangdong Institute of Safety Science and Tech. , Guangzhou 510620 , Guangdong , China
  • Received:2006-11-06 Online:2007-09-25 Published:2007-09-25
  • Contact: 周剑峰(1970-) ,男,博士生,主要从事过程装备安全可靠性及风险评价技术研究. E-mail:zhou_jianfeng@sohu. com
  • About author:周剑峰(1970-) ,男,博士生,主要从事过程装备安全可靠性及风险评价技术研究.
  • Supported by:

    广东省科技计划项目(2005B32401008 )

摘要: 针对危险源特别是重大危险源的分布式远程监控预警系统对预防重大事故、应急救援、管理决策等具有重要作用.文中基于多Agent 技术,提出了多Agent 纽(MAG) 的概念和模型,并分析其特点及与多Agent 系统(MAS) 的区别与联系.进而在危险源监控任务的基础上建立了危险源监控的MAG 功能模型,即把针对一个危险源的监控任务分解到监控Agent 、远程通信Agent 、特征提取Agent 、预警Agent 和界面Agent 等5 个Agent 中,它们自主运行并通过协作构成一个多Agent 纽共同完成监控任务,由多个分布的MAG构成分布式危险源监控系统.最后基于以上MAG 模型和分布式监控系统结构,建立了液化石油气储存装直的危险源监控原型系统,分析了该系统中实现的各个Agent 的主要功能.

关键词: 危险源, 监拉系统, 多Agent, 团队, 分布式系统

Abstract:

The distrihuted and remote monitoring and controlling system for hazard installations , especially for fatal hazard installations , is of great importance to the prevention of major accidents , the succor of people and properties in emergency and the management of hazard installations. In this paper , the concept and model of multi-Agent group (MAG) are first proposed hased on the multi-Agent technology. Then , the characteristics of the MAG model
and the differences hetween MAG and multi-Agent system (MAS) are discussed. Moreover , a MAG model for monitoring and controlling hazard installations is estahlished hased on task analyses , in which the tasks are decomposed into 5 Agents including the Monitoring and Control Agent , the Remote Communication Agent , the Feature Extraction Agent , the Early-Waming Analysis Agent and the User Interface Agent. The 5 Agents operate independently and they synergetically compose a multiple distrihuted MAGs system to perform distrihuted monitoring and controlling for hazard installations. Finally , a monitoring and controlling system for LPG storage is presented hased on the proposed MAG model and the architecture of the distrihuted monitoring and controlling system , and the main functions of the agents are also analyzed.

Key words: hazard installation, monitoring and controlling system, multi-Agent, group, distrihuted system