残缺判断信息下专家共识群决策理论、方法及应用研究

: This project studies the theory, methods and applications of consensus in group decision making with incomplete preference relations. Based on our research achievements on the priority methods of incomplete fuzzy preference relations and other uncertain decision making problems, learning from group consensus methods under complete preference relations, this project will construct the new group consensus frameworks for different types of incomplete preference relations. As the ordinal consistency of a preference relation is the minimum condition that a decision maker is not self-contradictory, this project will study the ordinal consistency of incomplete preference relations using the graph theory, and then develop the group consensus models based on ordinal consistency. This project will study the consensus methods for different types of incomplete preference relations, such as incomplete fuzzy preference relations, incomplete multiplicative preference relations, incomplete interval preference relations, incomplete triangular fuzzy preference relations, incomplete linguistic preference relations, and incomplete intuitionistic fuzzy preference relations, as well as their mixtures. In the process of group consensus, this project will study how to modify a single judgment of individual decision maker when the consensus degree is not within the predefined threshold, this aims to preserve the decision makers' original information as much as possible. Based on the above researches, this project will further study the judgment level for each decision maker, and develop a network platform based on the above group consensus methods. It would provide the auxiliary decision supports for the real decision making problems, such as unconventional water disaster emergency, foundation evaluations, etc.

本项目研究残缺判断信息下专家共识的群决策理论与方法及应用，以申请人在残缺模糊判断的排序方法及不确定决策方面的研究成果为基础，借鉴完全信息下的专家共识群决策方法，建立不同类型残缺判断信息下专家共识过程的框架；考虑到次序一致性是决策者个体判断一致性的基本条件，运用图论研究残缺判断的次序一致性，进而研究基于次序一致性的专家共识模型；分别研究不同类型残缺偏好信息下的专家共识决策方法，这些类型包括残缺模糊互补判断，互反判断，区间判断，三角模糊判断，语言判断和目前广泛关注的直觉模糊判断，以及基于上述不同偏好类型的混合残缺判断；在共识过程中，若专家共识度未达到预先定义的阈值时，研究调整个别专家个别判断信息的方法，以最大限度的保持专家原来的判断信息。在上述研究基础上，研究专家各自的判断水平；研究开发基于上述群体共识方法的网络研讨平台，为非常规突发水灾害事件、基金评审等现实决策问题提供辅助决策支持。

专家共识，即群体一致性，是群决策研究中的热点。专家群体达成一致的决策可以兼顾多方面的利益，目前，我国一些重大决策问题迫切需要群体决策，且专家之间基本达成一致。在决策过程中，决策者给出的是偏好信息。决策者对方案之间的两两优劣比较容易，这就是应用广泛的判断矩阵。然而，在进行决策时，当层次很多，因素复杂时，总判断量有时是很大的。很可能由于时间紧迫（如突发事件），决策者对某些比较判断缺少把握、不感兴趣、或者对某些敏感问题不想发表意见，称之为残缺判断。本项目研究了残缺判断情形下的专家共识决策理论与方法，取得了一系列富有创新性的研究成果，如（1）提出了残缺犹豫模糊互补偏好关系的概念，定义了残缺犹豫偏好关系的加性一致性和积型一致性的概念，然后提出了两个目标规划模型来对加性一致性和积型一致性分别导出权重向量。并且将这两个目标规划模型扩展到群体残缺犹豫偏好关系；（2）提出了犹豫模糊互补偏好关系的共识决策方法，并在水资源管理中进行了应用；（3）提出了加性一致性的犹豫二元语言偏好关系的群体共识模型；（4）提出了犹豫模糊偏好关系的动态权重调整的共识决策方法。项目成果发表在《Knowledge-Based Systesms》、《Applied Soft Computing》、《Information Sciences》等国际知名期刊上，SSCI和SCI收录论文（不含录用）28篇，ESI高被引论文2篇。本项目的研究成果受到国内外同行的高度关注并被广泛应用。