Methods Based On Credibility Theory For Fuzzy Multi-attribute Decision-making

As an important component of the modern decision theory, the multi-attributedecision-making has been widely used in many fields. However, as the complexityof decision-making problem and the limited nature of people's cognitive abilities,in some cases, the decision-making information is often given in a fuzzy form.In this dissertation, the fuzzy multi-attribute decision-making problems are dis-cussed. The detailed contents are described as follows:The problem of ranking fuzzy variables is discussed. The concepts of thefuzzy dominance degree between two fuzzy variables and the unit entropy of fuzzyvariable are defined. The fuzzy dominance degree measures the degree that onefuzzy variable dominates the other fuzzy variable. The unit entropy describesthe fuzziness that the unit expected value shares. The ranking criterion based onthe fuzzy dominance degree is proposed. The unit entropy criterion is presentedas well.The multi-attribute decision making problem is discussed, in which the at-tribute weights are unknown completely. The distance of between LR-type fuzzyvariables is proposed based on theα-pessimistic value and theα-optimistic val-ues of fuzzy variables. The distance is used to measure the difference amongthe attribute values. The programming model is established to determine theattribute weights based on the ideal of maximizing the deviation. Consideringthat the normalized attribute values are all benefit attribute values, the dis-tance method for multi-attribute decision-making is presented. The referencepoint dominance criterion is given, as a result that the membership degree ofthe vertex of the triangular fuzzy variable is 1. The method based on the fuzzydominance degree between fuzzy variables is proposed. The programming modelis established to determine the fuzzy positive idea solution and the fuzzy nega-tive ideal solution, in which the expected value and the entropy of the attributevalue are taken into accounts at the same time. The relative closeness degreeis improved. The example is given to illustrate that the different fuzzy positive idea solution and fuzzy negative ideal solution determined using different meth-ods may change the result of the alternatives. Furthermore, the multi-periodmulti-attribute decision-making problem is discussed.The multi-attribute decision-making problem is discussed, in which the at-tribute weights are known partly. According to different risk preferences of de-cision makers, the two-stage optimization models are proposed. Finally, basedon the expectations, entropy, unit entropy of the integrated attribute value, aswell as the deviation from the fuzzy positive ideal solution, the correspondingdecision-making methods are put forward.