【作者】 王翯华；

【导师】 方志耕；

【作者基本信息】 南京航空航天大学 ， 管理科学与工程， 2013， 博士

【摘要】 决策是人们为了达到某种目的或完成某种任务而进行的有意识、有选择的行动过程。在实际决策过程中，决策环境错综复杂、决策信息瞬息万变。为了尽可能做出科学合理的决策，在充分收集和利用信息数据的同时，必须依靠群体的智慧，强化决策技术与模型的研究。目前，国内外学者在确定性环境方面的决策技术和方法取得了很多成果，但是对不确定条件下、特别是自然语言环境下的群决策问题研究尚缺乏较为深入和系统的理论体系，相关研究处于起步和探索阶段。不确定条件下的语言评价信息群决策问题的系统研究既有迫切需要，也有其现实意义。论文紧密围绕语言评价信息决策环境，研究了面向语言评价信息的群体决策模型，主要创新工作如下：（1）研究了基于量变和质变的语言评价信息质量分析模型以及信息不完全情况下的信息估算模型。在复杂问题决策过程中，由于决策群体分布广泛、决策群体的结构和决策环境越来越复杂，评价决策信息日趋呈现出规模庞大、来源复杂、渠道多样而导致质量良莠不齐的特点。主要依据评价过程中语言评价信息的特点、事物发展规律及决策者判断质量的测度等因素，提出了解决基于量变和质变的不确定语言评价信息质量分析的新方法；建立了方案各阶段量变与质变的判定模型，以利于决策者对决策信息质量的充分挖掘；提出了多阶段语言评价信息下的决策者权重确定方法，提高了语言评价信息动态决策的有效性，进一步发展和完善了基于语言评价信息质量推理的群决策理论与方法。（2）研究了基于灰色关联度的多阶段语言评价信息集结模型。研究了多阶段语言评价信息的决策特点，结合被评价事物在各阶段决策中的发展与潜在趋势，提出了事物在评价阶段是否发生变异的判断方法；建立了基于灰色关联理论的多阶段语言评价信息问题的时间权重确定模型；根据群体差异最小思想，提出了多阶段语言评价信息的群集结方法。（3）研究了基于灰色关联度的大规模群决策集结模型。研究了群体规模较大情况下的群体聚类方法，建立了采用灰关联聚类的方法，将大规模群体进行适度分类，以降低大规模群体的协调难度；根据决策者所属群体类别差异，提出了类内评价指标权重确定方法；测算了各群体类别的综合偏好，提出了核心决策者确定方法；研究了群体类间权重确定模型，建立了基于群体类别意见偏差最小的群体偏好信息集结模型；提出了决策者判断一致度定义及决策者偏差协调方法。（4）基于案例分析的语言评价信息灰靶决策分类模型。针对比较常用的自然语言评价信息，依据自然语言和模糊数的对应关系，结合灰靶决策过程靶心设置的逆序问题，提出语言灰靶决策过程中的靶心设置及修正方法；以案例符合度最大为目标，考虑靶心距分类临界值约束和决策者对指标权重约束，建立了基于模糊靶心距离的多属性决策分类模型，优点是基于靶心距离模糊比较后，能了解各方案的偏差程度和总体偏差。（5）面向商用飞机供应商选择评价的方法应用。结合商用飞机供应商选择评价的特点，提出了基于语言评价信息的选择评价方法体系；运用语言评价信息的群决策方法对大型客机的供应商选择评价进行了实际研究，促进了决策方法和模型的实际应用，为解决该环境下的不确定语言信息决策做了有益的探索。更多还原

【Abstract】 Decision-making is a conscious and selective action process that people do to achieve a certainpurpose or to complete a certain task. In the actual decision-making process, the environment iscomplex and the information is rapidly changing. In order to make a scientific and reasonabledecision, the wisdom of group is more required, as well as improving decision-making technologywhen fully gathering and using information. At present, the domestic and overseas scholars haveobtained many achievements about decision-making techniques and methods of groupdecision-making in certainty condition. However, there is still a lack of theoretical system for underuncertainty group decision-making problems based on linguistic information. The relevant domesticresearch is at the stage of exploration currently. The systematic research on under uncertainty groupdecision-making problems based on linguistic information is urgently needed, also has its practicalsignificance. This paper researches on linguistic-information-oriented model about groupdecision-making. Main innovations of this paper are described as follows:(1) This paper has researched on the information quality analysis model of linguistic assessmentbased on quantitative and qualitative changes and information estimated model about incompleteinformation. In the decision-making process of complicated problems, the distribution and structure ofthe decision group and the decision environment become more and more complex, so the assessmentdecision-making information increasingly shows a spotty feature due to the huge scale, complexsources and diverse channels. Based on these factors, i.e. the characteristics of the linguisticassessment information, the law of things development and the judgment quality of decision makers,this paper proposes a new solution to quality analysis of uncertain linguistic assessment informationbased on quantitative and qualitative changes. In order to help decision makers excavating the qualityof decision-making information sufficiently, this paper also represent a judgment method ofquantitative and qualitative changes at different stages. What’s more, this paper raises a method todetermine the weight of decision makers under multi-stage assessment information, which enhancesthe effectiveness of dynamic decision-making about the linguistic assessment information andimproves the group decision-making theory and method based on linguistic assessment informationquality reasoning.(2) This paper has studied aggregation model of multi-stage linguistic assessment informationbased on grey incidence degree. With fully discussing the characteristics of multi-stage linguistic assessment information and combining the potential trends and development of evaluated things ineach decision-making stage, a judgment method is presented to analyze whether mutation happenedduring assessment. In this paper, a model of time weight determination based on grey correlationtheory is built about multi-stage group decision-making problems linguistic assessment information.According to minimizing group differences theory, this paper represents an aggregation method ofmulti-stage linguistic assessment information.(3) This paper proposes an aggregation model of large scale group decision-making based ongrey incidence degree. Studying large scale group cluster methods and combining the grey relevanceclustering method, this paper classifies the large scale group into reasonable groups to make it easierto coordinate the large group. According to the difference of decision-maker groups, a method isdiscussed in this paper to determine decision makers’ weights in respective group. Also, this papermeasures the general preferences for each group and presents a method to confirm the core decisionmakers. With the model of the weights determination among groups, a group preferences aggregationmodel is proposed based on minimizing group decision differences.(4) This paper has researched on linguistic assessment information grey target decision-makingclassify model based on case study. According to the correspondence between natural language andfuzzy number, using common natural linguistic assessment information and combining the inverseproblem of grey target decision-making clouts setting, this paper represents clouts setting and fixingmethod in linguistic grey target decision-making process. To maximize the conformity of the case,this paper builds a multi-attribute decision-making classify model based on fuzzy clouts distance,considering the critical value constraint of classification about the distance between two clouts and theindex weight constraint caused by decision makers. Because of the fuzzy comparison based on cloutsdistance, it is clear that this model can easily show the population deviation and the deviation of eachscheme.(5) This paper has applied the assessment method above into the selection assessment of largeaircraft supplier. According to the characteristics of the selection of large aircraft supplier, this paperpresents a selection assessment method system. What’s more, this paper researches on actual largeaircraft supplier selection assessment through the group decision-making method of linguisticassessment information. This promotes the development of the theory and practical application ofdecision-making methods and models. We are sure that it makes beneficial exploration on how tosolve uncertain linguistic information decision-making problems under this circumstance.更多还原