【作者】 刘明广；

【导师】 杜纲；

【作者基本信息】 天津大学 ， 管理科学与工程， 2007， 博士

【摘要】 随着决策群体规模和决策问题的复杂程度不断加大,研究复杂群决策问题已经成为群决策研究领域的重要课题。但目前的研究总体上缺乏系统的理论指导体系,已有的成果还很少,可以说这方面的研究还处于起步和探索阶段。本论文将基于复杂性科学的相关理论与方法,对复杂群体决策问题特别是其优化机理与方法进行深入分析和研究,主要工作和创新如下:第一,应用复杂性科学的相关理论,系统地分析了群决策系统的复杂性表现,结合涌现的特点与本质,提出了复杂群决策系统的涌现现象以及定义,并进一步探讨了复杂群决策系统涌现来源及其形成机理。第二,基于涌现的内涵与机理,提出了复杂群决策系统的协同指导思想、协同主导逻辑以及协同类型,并设计了基于涌现的多主体群决策支持系统协同框架。第三,从现实背景出发,分析了一类具有多层次结构的复杂群决策问题的典型特征之后,将主从优化思想与具有多层次结构组织的群决策问题相结合,提出了复杂群决策系统主从递阶协同优化模型的标准形式以及类型。第四,针对复杂群决策系统的主从递阶协同优化模型的求解难题,借鉴生物界协同进化的相关理论与思想,以差异演化算法为基础算法,提出了基于种群密度的协同差异演化算法,并设计了复杂群决策系统主从递阶协同优化模型的通用算法求解方案。第五,在提出了大群体带有语言变量的多准则复杂群决策问题的定义之后,建立了一种面向大群体带有语言变量的多准则协同优化方法,即直接通过测量各个体的方案准则集结值与群体方案准则理想解之间的距离来确定最终的群体方案排序。第六,针对决策群体对方案评价的不同偏好信息形式,在建立决策群体满意度函数的基础之上,提出了一种基于群体满意度最大的多形式信息协同优化模型。更多还原

【Abstract】 With the size of group decision-making growing and the complexity of decision-making increasing, studying group decision-making problem has become a critical issue in the research field of group decision-making. However, there have been no systemic theory guidelines estabilished in the research field as a whole and few achievements have been made. Thus the exploring in this field is on its initial stage. Analysis and research in the field of complex group decision-making problem and its optimization mechanism are conducted in this thesis based on the theory and methods of complex science. The major work and innovative ideas are as follows:Firstly, the complexity of group decision-making system is systematically analyzed applying the theory of complex science and integrating the characteristic and nature of emergence. The emergence phenomenon and its definition of the complex group decision-making system were presented, and further research in the origin of emergence and its developing mechanism of the complex group decision-making system is conducted.Secondly, collaboration-guided theory, collaboration-dominated logic and the collaboration styles of the complex group decision-making system are presented herein based on the content and mechanism of emergence; and the collaborative frame of the multi-agent group decision-making support system is designed grounded on emergence.Thirdly, typical features of a kind of complex group decision-making issues with multi-layer structure are practically analyzed, and the standard forms and types of the leader-follower collaborative optimization model of the complex group decision-making system are built integrating the ideology of leader-follower optimization and the practical group decision-making issues with multilayer-structured organization.Fourthly, as to the difficult solution of the leader-follower collaborative optimization model of the complex group decision-making system, the collaborative-differential evolution algorithm is established based on differential evolution algorithm and species group density, referring to the theory and idea of collaborative evolution of the nature.Fifthly, the definition of the multi-disciplined complicated group decision-making problem of the large group with linguistic variance is presented, and a multi-disciplined collaborative optimization method upon the large group with linguistic variance is subsequently constructed, which is to measure the distance between the value of each individual plan discipline set and that of the group discipline ideal solution.Sixthly, as the decision-making group exhibited preference to information patterns in the plan appraisal, the thesis builds a satisfaction function of the decision-making group and sequentially a multi-patterned information collaborative optimization with maximum group satisfaction.更多还原