【作者】 左春荣；

【导师】 杨善林；

【作者基本信息】 合肥工业大学 ， 管理科学与工程， 2008， 博士

【摘要】 决策无处不在,无时不有。在制定经济与社会发展战略的过程中和宏观经济运行管理中有大量的决策科学问题,在企业经营管理过程中和复杂新产品开发、制造及其使用维护过程中也有大量的决策科学问题。随着经济社会的高速发展,国际竞争日趋激烈,使得决策过程充满了随机性、模糊性、多样性、博弈性和群体性,对决策的科学性提出了更高的要求,因此必须深入研究能够处理不确定性问题的群决策理论与方法。本文以定性群决策问题为研究背景,在研究群决策推理机制和基于定性推理的智能决策方法的基础上,深入研究了群决策中的定性推理、基于定性推理的群决策机制、群决策的结果集结和过程集结、基于Multi-agent的定性推理群决策支持系统体系结构、基于Multi-agent的群决策支持系统协调器的结构和协调器的融合策略等有关基于定性推理的群决策方法和群决策支持系统方面的基本理论问题,最后以轿车整车概念设计中的定性群决策问题为例,研究了概念设计智能决策支持系统,将决策仿真过程引入到车型造型设计系统中,并进行了决策仿真过程设计。本文的主要研究内容和创新性工作如下:(1)在分析定性推理理论、Agent技术和群决策方法的研究现状的基础上,重点研究了群决策中的集结问题。(2)研究了复杂定性群决策中的由定性变量、定量变量构成的定性定量集成决策模式,提出了一种基于定性推理的复杂定性群决策问题处理过程和相应的处理方法。研究了定性变量与定量变量之间的转换方法、用于定性群决策问题的数据库、模型库、方法库和知识库等决策信息资源库及其管理系统的结构以及复杂定性群决策问题的分解和求解过程的协调机制。(3)研究了基于定性推理的群决策机制,提出了基于定性推理理论的群决策结果集结方法和过程集结方法。通过增加判断约束和赋值约束,设计了结果集结环境下改进的QSIM算法;用定性状态转换关系描述群决策过程中决策者之间由于外界环境、相互交流等因素对决策者决策行为的影响规则,设计了群决策过程集结中决策者之间的交流过程和评价过程算法。(4)研究了基于Multi-agent的定性推理群决策支持系统的功能组成、运行机制和多Agent之间的协调控制策略,提出了一个基于Multi-agent的群决策支持系统结构,设计了基于定性推理的交互Agent、结果集结Agent、过程集结Agent和协调控制Agent的结构模型,并深入分析了它们的运行过程。(5)以轿车整车概念设计为例,研究了基于Multi-agent的概念设计中的群决策支持系统结构,给出了系统中各类信息资源库的构成;针对概念设计中的车型造型设计问题,运用设计结构矩阵构建了仿真模型,给出了相应的仿真算法,通过该仿真过程能够估算车型造型设计阶段的执行时间;基于造型件的设计思想,构建了基于Multi-agent的仿真系统基本框架,设计了系统中各类Agent的功能,该仿真系统能够较好地支持设计者之间的相互交流和造型设计方案的形成。更多还原

【Abstract】 Decision making is ubiquitous at all time. There exist lots of scientific decision problems in the process of economic and social development strategy making and macro-economy operation management. There also exist large amount of such problems in the process of enterprises’ operation management and complex new products development, manufacture, operation and maintenance.The growing international competition under the rapid development of economy and society brings decision process randomicity, ambiguity, diversity, game playing and groupment, and requires higher scientificity for decision making. Therefore, group decision theories and methods for uncertain problems should be given intensive study.Aiming at qualitative group decision issue, this dissertation mainly discusses some basic academic problems in group decision making method and group decision support system based on qualitative reasoning, such as qualitative reasoning in group decision, group decision mechanism based on qualitative reasoning, result aggregation and process aggregation of group decision, and the architecture of qualitative reasoning group decision support system based on multi-agent, the structure of coordinators and the fusing policy of coordinators in group decision support system based on multi-agent. Taking conceptual design of car system as instance, the intelligent group decision support system of conceptual design is studied, and decision simulation process is introduced to the car modeling design system and an algorithm of decision simulation process is designed. The details are as follows.(1) The aggregation problems in group decision making are discussed based the analysis of qualitative reasoning theory, agent technology and group decision making methods.(2) Quantitative variables and qualitative variables integrated group decision making mode is discussed. A kind of processing procedure and method for complex qualitative group decision making based on qualitative reasoning is presented. The conversion method between qualitative variables and quantitative variables is discussed. The function of decision information resource bases consisting of database, model base, method base and knowledge base is analyzed. The decomposition of complex group decision problem and the coordination mechanism of problem solving procedure are also discussed.(3) Group decision making mechanism based on qualitative reasoning is discussed. A result aggregation method and a process aggregation method for group decision making based on qualitative reasoning are presented. An improved QSIM algorithm for result aggregation is designed via adding a judgment constraint and an assignment constraint. The influence rules of decision behavior of decision makers caused by outer circumstance and communication are described with qualitative state conversion relation, and an algorithm for communication process and evaluation process among decision makers in group decision making process aggregation is designed.(4) The function composition, operation mechanism and the coordination policy among agents of qualitative reasoning group decision support system based on multi-agent are discussed. A structure of group decision support system based on multi-agent is presented. The structure models of interaction agent, result aggregation agent, process agent and coordination agent are designed, and their operation process is also analyzed. .(5) Taking the conceptual design of car system as instance, the structure of the group decision support system for conceptual design based multi-agent is discussed. The composition of the decision information resources is described. A simulation model based on design structure matrix is constructed for modeling design of car system, and the corresponding simulation algorithm is designed. The execution time of modeling design of car system can be estimated via this simulation process. On the basis of modeling parts design idea, a basic frame of simulation system based on multi-agent for modeling design of car system is designed. The function of various agents in the simulation system is described. This simulation system can support the communication among designers and the formation of new schemes of modeling design effectively.更多还原