基于Multi-Agent的复杂工业过程控制系统及其在PROFIBUS上实现的研究
|
摘要
随着信息技术的飞速发展，工业过程控制领域正孕育着深刻的变革。控制要求的不断提高，控制程度的愈加复杂使长久以来在生产中运用的单一的经典控制理论和技术已经不能满足系统的需求，现代化工业过程控制系统迫切地需要高级理论的支撑。与理论发展相对滞后形成对比的是，自动化技术工具的蓬勃发展，在经历气动仪表控制、集中式数字控制系统到集散式控制系统后形成了新一代的网络集成式全分布的现场总线控制系统。本文研究的就是如何将工业控制领域中任务的需要、理论的开拓与技术工具和手段的进展结合起来，三者共同促进工业过程控制系统的发展。
Multi-Agent理论是分布式人工智能领域的重要分支之一。因为具有较强的自主性、分布性、协调性、拓展性及逼真的表现复杂系统、解决可分解的复杂问题的能力，该理论一直是近年来的研究热点。本文从复杂工业过程控制生产发展的需要出发，将Multi-Agent理论引入传统的工业过程控制领域，提出了建立基于Multi-Agent理论的复杂工业过程控制系统的命题，并根据现场总线系统的特点对系统在现场总线上实现进行了探讨。
首先通过对现代化工业过程控制系统特点的分析，对Multi-Agent理论和现场总线的简单介绍，指出进行本课题的必要性和可行性。进一步介绍广义Multi-Agent理论的基础上探讨工业过程控制领域通用的基于Multi-Agent的复杂工业控制系统的架构，定义了符合工业过程控制系统特点的个体Agent模型和个体Agent的行为规则。构建了两种不同的Agent群体社会体系结构：分散式体系结构和混合式体系结构。根据不同的群体体系结构提出了不同的协作机制：联合行动求解机制和协调仲裁管理机制。接着从技术特点到结构特点对现场总线系统和基于Multi-Agent的工业过程控制系统进行了细致的比较分析，指出两者从思想方法到解决问题的模式有惊人的相似性，进一步论证了前者是后者的天然实现平台。制定了实现平台的选型原则并选取
基于Multi一Agent的复杂工业过程控制系统及其在pR〔)FIBuS上实现的研究
PROFIBUS总线作为本文研究的系统的实现平台。选择具有代表性的温度一
液位多变量控制对象具体构建了基于Multi一Agellt的多变量控制系统。提出了
新型的系统控制需求，根据需求对Ag饥:进行了功能的划分和定义，并对系
统中主要功能的Agent进行了算法的选取和仿真研究。最后给出了在
pRoFBus上实现该基于MUlti一翰。t理论的多变量控制系统的软、硬件总体
方案和主要Agent的实现方法。
本文的研究理论与生产需求相结合，具有较强的理论与实践意义。将相
关领域的先进理论引入工业过程控制领域，为自动控制理论的发展开辟了一
条新路，也为复杂工业控制理论找到了一个新的出发点。
陈玲(控制理论与控制工程)
指导教授:丁纪凯
With the rapid development of information technology, a deep change has taken placed in the industry process control field. The simple control theory used in this field for years can't meet the new requirement of the system any more due to the higher control target, and a more advanced theory need to be found to support the system with higher complexity, hi contrast with the slow progress of the control theory, the automatic control tool experienced the evolution of pneumatic meter control, centralize data control system and distributed control system has made a great achievement, that is the new net integrated and distributed control system-field bus control system. The purpose of this research is to find a way in which the control requirements can be combined with the progress of the control theory and control tools in order to promote the development of industry process control system.
The theory of Multi- Agent, which has been quickly developed in recent years, Hs one of the most important branches of the DAI. Because of its abilities of adaptively, Autonomy and cooperation, the theory has superiority in process of solving complex problems and simulating complex system, hi this paper, a complex industry process control system based on Multi-Agent theory has been proposed and its realization on the field bus system also has been mentioned.
At beginning of the paper, the necessaries and possibilities of the research were discussed based on the analysis of the characteristic of modern industry process control system. Then the frame of the complex industry process control system based on Multi-Agent theory for a broad senesce was designed. The single Agent model and the behavior rule for it were defined. Two different Agent colony structure, the distracted and hybrid structure, then were given in the essay.
A compare was made between the complex industry process control system based on Multi-Agent theory and FCS from the technical characteristic to architectural characteristic in detail and the conclusion that FCS can act as the a generic system platform to realize the complex system was drawn based on the
compare. A strategy on how to meet various field buses' challenge was also given. A cording to the strategy,, PROFIBUS was chosen to be the realization platform in tliis research.
As a representational case of the complex industry control system, a multi-variable system whose controlled objects were the temperature and level of the water was built up based on Multi-Agent theory. The system was divided into several different Agents with different control functions to accomplish the new control assignments. The arithmetic simulation research of major Agent was done after function was divided. At the last, both of the hardware and software plan how to realize the multi-variable system on PROFIBUS was given.
The essay combines the theory research and practice together. Brought advanced theory from other related field into industry process control field, the paper pointed out a new way for the research of automatic control theory.
ChenLing(Control Theory and Control Engineering)
Supervised by Ding Jikai
Multi-Agent理论是分布式人工智能领域的重要分支之一。因为具有较强的自主性、分布性、协调性、拓展性及逼真的表现复杂系统、解决可分解的复杂问题的能力，该理论一直是近年来的研究热点。本文从复杂工业过程控制生产发展的需要出发，将Multi-Agent理论引入传统的工业过程控制领域，提出了建立基于Multi-Agent理论的复杂工业过程控制系统的命题，并根据现场总线系统的特点对系统在现场总线上实现进行了探讨。
首先通过对现代化工业过程控制系统特点的分析，对Multi-Agent理论和现场总线的简单介绍，指出进行本课题的必要性和可行性。进一步介绍广义Multi-Agent理论的基础上探讨工业过程控制领域通用的基于Multi-Agent的复杂工业控制系统的架构，定义了符合工业过程控制系统特点的个体Agent模型和个体Agent的行为规则。构建了两种不同的Agent群体社会体系结构：分散式体系结构和混合式体系结构。根据不同的群体体系结构提出了不同的协作机制：联合行动求解机制和协调仲裁管理机制。接着从技术特点到结构特点对现场总线系统和基于Multi-Agent的工业过程控制系统进行了细致的比较分析，指出两者从思想方法到解决问题的模式有惊人的相似性，进一步论证了前者是后者的天然实现平台。制定了实现平台的选型原则并选取
基于Multi一Agent的复杂工业过程控制系统及其在pR〔)FIBuS上实现的研究
PROFIBUS总线作为本文研究的系统的实现平台。选择具有代表性的温度一
液位多变量控制对象具体构建了基于Multi一Agellt的多变量控制系统。提出了
新型的系统控制需求，根据需求对Ag饥:进行了功能的划分和定义，并对系
统中主要功能的Agent进行了算法的选取和仿真研究。最后给出了在
pRoFBus上实现该基于MUlti一翰。t理论的多变量控制系统的软、硬件总体
方案和主要Agent的实现方法。
本文的研究理论与生产需求相结合，具有较强的理论与实践意义。将相
关领域的先进理论引入工业过程控制领域，为自动控制理论的发展开辟了一
条新路，也为复杂工业控制理论找到了一个新的出发点。
陈玲(控制理论与控制工程)
指导教授:丁纪凯
With the rapid development of information technology, a deep change has taken placed in the industry process control field. The simple control theory used in this field for years can't meet the new requirement of the system any more due to the higher control target, and a more advanced theory need to be found to support the system with higher complexity, hi contrast with the slow progress of the control theory, the automatic control tool experienced the evolution of pneumatic meter control, centralize data control system and distributed control system has made a great achievement, that is the new net integrated and distributed control system-field bus control system. The purpose of this research is to find a way in which the control requirements can be combined with the progress of the control theory and control tools in order to promote the development of industry process control system.
The theory of Multi- Agent, which has been quickly developed in recent years, Hs one of the most important branches of the DAI. Because of its abilities of adaptively, Autonomy and cooperation, the theory has superiority in process of solving complex problems and simulating complex system, hi this paper, a complex industry process control system based on Multi-Agent theory has been proposed and its realization on the field bus system also has been mentioned.
At beginning of the paper, the necessaries and possibilities of the research were discussed based on the analysis of the characteristic of modern industry process control system. Then the frame of the complex industry process control system based on Multi-Agent theory for a broad senesce was designed. The single Agent model and the behavior rule for it were defined. Two different Agent colony structure, the distracted and hybrid structure, then were given in the essay.
A compare was made between the complex industry process control system based on Multi-Agent theory and FCS from the technical characteristic to architectural characteristic in detail and the conclusion that FCS can act as the a generic system platform to realize the complex system was drawn based on the
compare. A strategy on how to meet various field buses' challenge was also given. A cording to the strategy,, PROFIBUS was chosen to be the realization platform in tliis research.
As a representational case of the complex industry control system, a multi-variable system whose controlled objects were the temperature and level of the water was built up based on Multi-Agent theory. The system was divided into several different Agents with different control functions to accomplish the new control assignments. The arithmetic simulation research of major Agent was done after function was divided. At the last, both of the hardware and software plan how to realize the multi-variable system on PROFIBUS was given.
The essay combines the theory research and practice together. Brought advanced theory from other related field into industry process control field, the paper pointed out a new way for the research of automatic control theory.
ChenLing(Control Theory and Control Engineering)
Supervised by Ding Jikai
引文
[1]蒋藯孙等．大型复杂工业生产过程综合自动化的理论和方法研究．“工业综合自动化”研讨会论文集，1994，5，36-41
[2]冯纯伯．复杂工业过程的控制问题．控制理论与应用．1992，9(3)303-305
[3]戴冠中，徐乃平．计算机对自动控制理论的挑战．控制理论与应用．1992，9(3)312-319
[4]Eugenio Oliveira: Towards Intelligent Robotic Assemblies. IEA/AIE1992, 361-370
[5]欧海涛等．基于RMM和贝叶斯学习的城市交通Multi-Agent系统．控制与决策．2001，16(3)，291-295
[6]N V Findler. Distributed control of collaborating and learning expert systems for street traffic signals[A].IFAC Distributed Intelligence Systems[C]. Pergamon Press, 1991,125-130
[7]B Burmeister, ect. Application of multi-agent systems in traffic and transportation [J]. IEEE Trans on Software Engineering, 1997, 144(1),51-60
[8]C Ramos. Architecture and a negotiation protocol the dynamic scheduling of manufacturing systems[A]. IEEE Int Conf on Robotics and Automation[C]. USA, 1994,(4) 8-13
[9]F Sprumont, etc. AMACOIA: A multiagent system for designing flexible assembly line[J].Applied Artificial Intelligence, 1997, 11(6),573-589
[10]刘金琨，尔联洁．多智能体技术应用综述．控制与决策．2001，16(2)，133-139
[11]阳宪惠 现场总线技术及其应用 北京：清华大学出版社 1999
[12]金敏等．实时分布式Multi-Agent控制理论的理想实现平台——现场总线控制系统．信息与控制．2001，30(4)，322-327
[13]D.M. lane Distributed Problem Solving and Real-time Mechanisms in Robot Architectures. Engng.Aplic.Artif.Intell., 1994,7(2):105-107
[14]W Kohu, A Nerode .Multiple Agent Autonomous Hybrid Control Systems[A]. Proc.of the IEEE Conf on Decision and Control[C]. Tucson, 1992.16-18
[15]Hewitt C, Inman J. DAI Betwixt and Between: From Intelligent Agents to Open Systems Science. IEEE Trans SMC, 1991, 21(6),1409-1419
[16]赵龙文，侯义斌．Agent的概念模型及其应用技术[J]．计算机工程与科学，2000，(6)：77-78．
[17]高波，费奇，陈学广．Agent交互层次模型[J]．计算机科学，2001，(8)：105-106
[18]董红斌，王建华．多Agent技术研究[J]．计算机应用研究，1999，(10)：29-30
[19]蒙祖强，蔡自兴．基于TS模型的Multi-agent系统建模及其稳定性分析．计算机技术与自动化．2002，(21)1，15-18
[20]A.J.N. van Breemen and T.J.A. de Vries, "Design and Implementation of a Room
Thermostat -sing an Agent-Based Approach," Control Engineering Practice, vol. 9, no. 3, pp.233-248, 2001.
[21]A Mehra, C Virginior .A Adaptive Process Control For A Plating Line Using Intelligent[A]. Int. Conf on Multiagent System, IXMAS' 98[C]. Paris,1998 76-82
[22]V Sahasabudhe .A Multiagent Control System Frame Work For Smart Structures[A].AIAA[C], USA 1998.4202-4215
[23]G.Hartvigsen and D.Johansen. Cooperation in a Distributed Artificial Intelligence Environmet-the Strorm-Cast application. Eng.Appli.of AI. 1990,(3), 229-237
[24]C.Seradas,H.Coelho and G.Gaspar. Communication Knowledge system:Part B. Big Talk among Small System. Appli. of AI. (1),233-260
[25]C.Zhang .Cooperation under Uncertainty in Distributed Expert System. AI. 1992 56, 21-69
[27]范玉顺，曹军威 多代理系统理论、方法与应用 北京靖华大学出版社2002．5 57-59
[28]R.Power. Mutual Intention. Journal for the Theory of Social Behaviour. 1994,14,85-105
[29]林强，阳宪惠 基金会现场总线测试技术 仪器仪表标准化与计量2000 35-8
[30]邬宽明．CAN总线原理和应用系统设计．北京：北京航空航天大学出版社，1992
[31]陈懿LonWorks技术为核心的现场控制系统，工业控制计算机1998 5 35-40
[32]夏德海 现场总线的现状及其应用 (三) 中国仪器仪表1998．3，1-4
[33]蔡忠勇，高涵。现场总线发展概况与DeviceNet进入中国 低压电器2000 2 59-61
[34]杨家强 冯晓生等PROFIBUS现场总线技术特点浅析 仪器仪表标准化与计量2002．3 29-32
[35]夏德海 现场总线的现状及其应用(二)中国仪器仪表1998．2，1-4
[36]唐济扬．PROFIBUS技术[J]．自动化博览，2001，18(4)38-40
[37]运用PROFIBUS-DP实现分散自动化 中国机电一体化技术应用协会
[38]SIMATIC Programming with STEP7 V5.1
[39]SIMATIC WINCC manual Version5.0
[40]万百五，黄正良 大工业过程计算机在线稳态优化控制：北京：科学出版社1998 1-5
[41]蒋藯孙，俞金寿 过程控制工程 北京：中国石化出版社1999．3
[42]彭志红，蔡自兴．基于遗传算法的非线性系统状态空间辨识 信息与控制2001，30(4) 289-291
[43]薛定宇 反馈控制系统设计与分析 北京：清华大学出版社2000．4 250-260
[44]王永初 仪表系统闭环整定方法 成都：四川科学技术出版社1984
[2]冯纯伯．复杂工业过程的控制问题．控制理论与应用．1992，9(3)303-305
[3]戴冠中，徐乃平．计算机对自动控制理论的挑战．控制理论与应用．1992，9(3)312-319
[4]Eugenio Oliveira: Towards Intelligent Robotic Assemblies. IEA/AIE1992, 361-370
[5]欧海涛等．基于RMM和贝叶斯学习的城市交通Multi-Agent系统．控制与决策．2001，16(3)，291-295
[6]N V Findler. Distributed control of collaborating and learning expert systems for street traffic signals[A].IFAC Distributed Intelligence Systems[C]. Pergamon Press, 1991,125-130
[7]B Burmeister, ect. Application of multi-agent systems in traffic and transportation [J]. IEEE Trans on Software Engineering, 1997, 144(1),51-60
[8]C Ramos. Architecture and a negotiation protocol the dynamic scheduling of manufacturing systems[A]. IEEE Int Conf on Robotics and Automation[C]. USA, 1994,(4) 8-13
[9]F Sprumont, etc. AMACOIA: A multiagent system for designing flexible assembly line[J].Applied Artificial Intelligence, 1997, 11(6),573-589
[10]刘金琨，尔联洁．多智能体技术应用综述．控制与决策．2001，16(2)，133-139
[11]阳宪惠 现场总线技术及其应用 北京：清华大学出版社 1999
[12]金敏等．实时分布式Multi-Agent控制理论的理想实现平台——现场总线控制系统．信息与控制．2001，30(4)，322-327
[13]D.M. lane Distributed Problem Solving and Real-time Mechanisms in Robot Architectures. Engng.Aplic.Artif.Intell., 1994,7(2):105-107
[14]W Kohu, A Nerode .Multiple Agent Autonomous Hybrid Control Systems[A]. Proc.of the IEEE Conf on Decision and Control[C]. Tucson, 1992.16-18
[15]Hewitt C, Inman J. DAI Betwixt and Between: From Intelligent Agents to Open Systems Science. IEEE Trans SMC, 1991, 21(6),1409-1419
[16]赵龙文，侯义斌．Agent的概念模型及其应用技术[J]．计算机工程与科学，2000，(6)：77-78．
[17]高波，费奇，陈学广．Agent交互层次模型[J]．计算机科学，2001，(8)：105-106
[18]董红斌，王建华．多Agent技术研究[J]．计算机应用研究，1999，(10)：29-30
[19]蒙祖强，蔡自兴．基于TS模型的Multi-agent系统建模及其稳定性分析．计算机技术与自动化．2002，(21)1，15-18
[20]A.J.N. van Breemen and T.J.A. de Vries, "Design and Implementation of a Room
Thermostat -sing an Agent-Based Approach," Control Engineering Practice, vol. 9, no. 3, pp.233-248, 2001.
[21]A Mehra, C Virginior .A Adaptive Process Control For A Plating Line Using Intelligent[A]. Int. Conf on Multiagent System, IXMAS' 98[C]. Paris,1998 76-82
[22]V Sahasabudhe .A Multiagent Control System Frame Work For Smart Structures[A].AIAA[C], USA 1998.4202-4215
[23]G.Hartvigsen and D.Johansen. Cooperation in a Distributed Artificial Intelligence Environmet-the Strorm-Cast application. Eng.Appli.of AI. 1990,(3), 229-237
[24]C.Seradas,H.Coelho and G.Gaspar. Communication Knowledge system:Part B. Big Talk among Small System. Appli. of AI. (1),233-260
[25]C.Zhang .Cooperation under Uncertainty in Distributed Expert System. AI. 1992 56, 21-69
[27]范玉顺，曹军威 多代理系统理论、方法与应用 北京靖华大学出版社2002．5 57-59
[28]R.Power. Mutual Intention. Journal for the Theory of Social Behaviour. 1994,14,85-105
[29]林强，阳宪惠 基金会现场总线测试技术 仪器仪表标准化与计量2000 35-8
[30]邬宽明．CAN总线原理和应用系统设计．北京：北京航空航天大学出版社，1992
[31]陈懿LonWorks技术为核心的现场控制系统，工业控制计算机1998 5 35-40
[32]夏德海 现场总线的现状及其应用 (三) 中国仪器仪表1998．3，1-4
[33]蔡忠勇，高涵。现场总线发展概况与DeviceNet进入中国 低压电器2000 2 59-61
[34]杨家强 冯晓生等PROFIBUS现场总线技术特点浅析 仪器仪表标准化与计量2002．3 29-32
[35]夏德海 现场总线的现状及其应用(二)中国仪器仪表1998．2，1-4
[36]唐济扬．PROFIBUS技术[J]．自动化博览，2001，18(4)38-40
[37]运用PROFIBUS-DP实现分散自动化 中国机电一体化技术应用协会
[38]SIMATIC Programming with STEP7 V5.1
[39]SIMATIC WINCC manual Version5.0
[40]万百五，黄正良 大工业过程计算机在线稳态优化控制：北京：科学出版社1998 1-5
[41]蒋藯孙，俞金寿 过程控制工程 北京：中国石化出版社1999．3
[42]彭志红，蔡自兴．基于遗传算法的非线性系统状态空间辨识 信息与控制2001，30(4) 289-291
[43]薛定宇 反馈控制系统设计与分析 北京：清华大学出版社2000．4 250-260
[44]王永初 仪表系统闭环整定方法 成都：四川科学技术出版社1984