【作者】 郭凯红；

【导师】 李文立；

【作者基本信息】 大连理工大学 ， 管理科学与工程， 2011， 博士

【摘要】 多属性群决策是现代决策科学的重要组成部分,长期以来是决策分析领域的前沿。信息不确定是实际多属性群决策问题的基本特征之一。在群决策环境下,如何有效实现不确定信息的集成与决策要素(包括属性值、属性权重及决策者权重)的获取具有重要的理论意义和现实意义。本论文以提高不确定群决策效率和质量为目标,按照“获取决策要素→集结决策信息→对比分析结果→提炼一般方法”的分析思路,引入不确定信息处理和管理决策分析等相关理论与方法,重点研究了群决策环境下不确定信息的集成规则和决策要素的获取方法,目的在于进一步丰富和完善不确定多属性决策理论与方法,增强复杂群体决策算法的实用性和灵活性。本论文的主要研究内容包括：(1)群决策环境下不确定信息集成规则及最少点决策模型研究。为更有效地实现多个决策者不同意见的融合,分别提出基于证据可信度和基于证据交叉融合策略的D-S证据理论合成方法,所提方法更有效地利用证据的可信度、期望支持度、交叉融合度等全局信息,确定了更理想的冲突分配策略,在证据一致和高度冲突的情况下均表现出良好的适应性；为进一步从群决策意见中获得有效的判决结果,深入研究最少点决策模型,系统解决了该模型中若干关键理论问题,使之成为一种实用化的决策方法。(2)群决策环境下权重信息获取方法研究。针对决策者权重及属性权重难以确定的问题,提出一系列在权重信息未知情况下属性值分别以不同模糊数据类型表示的不确定群决策方法。该系列方法基于决策者及不确定决策矩阵自身的特性,利用属性熵权、距离测度和模糊变换获取关键权重信息,更好地保证了不确定群决策的效率和质量。(3)群决策环境下基于态度的直觉模糊属性值构建方法研究。针对不确定混合属性值难以处理的问题,提出一种基于决策态度的直觉模糊属性值构建方法,目的是把不确定混合属性值统一转换成符合决策者态度的直觉模糊值。该方法充分考虑并形式化决策者的态度及偏好,在避免原始决策信息的损失和扭曲的同时,还可以根据不同的决策态度构造不同的直觉模糊属性值,在实际应用中具有更大的灵活性。本研究面向不确定群决策,为解决该环境下的不确定信息集成和决策要素获取这两个热点和难点问题做了有益探索,最后总结以上所提出的各种相关方法,给出针对权重信息未知情况下不确定群问题的一般方法。本研究成果进一步丰富和完善了不确定信息决策理论与方法,在经济、管理和军事等诸多领域具有广阔的应用前景。更多还原

【Abstract】 Multiple attribute group decision making, as an important component of modern decision science, has long been the forefront in the decision analysis filed, and uncertain information is one of its basic features in practice. Therefore, how to effectively combine uncertain information and acquire decision factors (including attribute values and weight information) under group decision making environment becomes an interesting and important research topic with theoretical and practical significance.To improve the efficiency and quality of group decision making under uncertainty, this dissertation, with an idea of "decision factors acquisition→decision information combination→results comparative analysis→general method extraction", employs some theories and methods related to uncertain information processing as well as management decision analysis to make an in-depth study on combination rules for uncertain information and approaches to decision factors under group decision making environment, aiming to further enrich and improve the theories and methods for multiple attribute decision making under uncertainty, and to greatly enhance the practicality and flexibility of complex group decision making algorithms. Specifically, the main contents of this dissertation include:(1) Study on combination rules for uncertain information and on the decision model based on least point’s principle under group decision making environment. For effective fusion of individual opinions of multiple decision makers, two different combination rules of evidence based on different strategy are presented, respectively. The two methods more effectively make use of global information such as the reliability, expected support, and the degree of cross merging between bodies of evidence, so as to determine the more satisfactory allocation strategy of evidential conflict, and therefore it has good adaptability to consistent or conflicting evidences, characterized by faster convergence rate and higher reliability. In order to obtain a wise decision result from aggregated opinions, the dissertation has an in-depth investigation on the decision model based on least point’s principle, and successfully solves several key theoretical issues remaining in the model, making the model a practical decision making method with further perfection in theory.(2) Study on approaches to the weight information under group decision making environment. In view of the difficulty in determining decision makers’weights and attribute weights, a series of methods for group decision making under uncertainty have been proposed in such situation where attribute values are respectively expressed as exact values, intervals, intuitionistic fuzzy values (IFVs) and interval-valued intuitionistic fuzzy values (IVIFVs) with totally unknown weight information. These methods are based on characteristics of decision makers and uncertain decision matrices themselves, and employ such theories as attribute entropy, distance measure and fuzzy transformation for key weight information and therefore can better guarantee the efficiency and quality of group decision making under uncertainty.(3) Study on an attitudinal-based method for constructing intuitionistic fuzzy information under group decision making environment. For the difficulty in handling hybrid attribute values, the dissertation develops an attitudinal-based method for constructing intuitionistic fuzzy information in attempting to convert uncertain hybrid attribute values into unified intuitionistic fuzzy values corresponding with a decision maker’s attitude which are relatively easy to handle. The presented method fully considers and formalizes a decision maker’s attitude as well as preference, and can construct intuitionistic fuzzy values corresponding with a decision maker’s attitude while avoiding the loss and distortion of original decision information, and therefore it has greater flexibility in practical applications.Oriented to group decision making under uncertainty, this research makes a useful exploration on the two difficult focuses under group decision making environment, i.e. combination of uncertain information and acquisition of decision factors. Finally, the dissertation summarizes the related methods proposed above to present a general solution to group decision making under uncertainty with totally unknown weight information. The achievements of this study further enrich and improve the existing decision making theories and methods, and have broad application prospects in many fields such as economics, management, and the military, etc.更多还原