Research On Risk-averse Multi-attribute Group Decision Making Methods For Hybrid Data

With the rapid development of information technology,the features of descriptive fuzziness,structural complexity and data scale have become increasingly prominent in realistic decision making problems,and complex decision making problems are constantly emerging,which have brought enormous challenges for traditional decision making models and methods.Facing with decision making problems with uncertain evaluation information and complex decision data,decision makers usually make analysis for decision alternatives from correlative and conflicting attributes,thus the research on multi-attribute decision making is emerged.Meanwhile,managers consider the weakness of inadequate individual cognitive capabilities for individual decision making modes,they usually form a decision group to obtain final decision conclusions,thus the research on group decision making is emerged.Moreover,decision makers combine multi-attribute decision making with group decision making to construct an efficient information fusion and analysis mechanism to get the preference order of the whole decision group,which forms the research of multi-attribute group decision making,and it has been successfully applied in many practical fields such as factory location,production planning,quality control,inventory management,etc.In multi-attribute group decision making,due to the influence of the decision-maker's different cognitive level and decision behavior characteristics,the description information of the decision object is usually represented by different granularity data types such as exact numbers,interval numbers and intuitionistic fuzzy numbers,so as to increase the practicability and the flexibility of evaluation systems.For the sake of taking advantages of several data types when expressing information,decision makers usually unite several data types to form decision matrices,and further develop hybrid multi-attribute group decision making.Nowadays,decision makers usually own risk-averse behavior characteristics when confronting with vital decisions,thus conducting decisions analysis with robust characteristics is necessary.In light of the above-statement,how to introduce risk-averse behavior characteristics into decision making methods has become a key issue of group decision making theories and methods.The thesis systematically studies risk-averse multi-attribute group decision making for hybrid data,starts from risk-averse behavior characteristics,utilizes exact numbers,interval numbers and intuitionistic numbers to form decision matrices,objectively explore acquisition methods of attribute weights and expert weights,and further constructs robust decision conclusions for hybrid multi-attribute group decision making via MULTIMOORA(Multi-Objective Optimization by Ratio Analysis plus the full MULTIplicative form).The main research contents of the thesis include the following three aspects:(1)The research on hybrid multi-attribute group decision making based on risk-averse decision making behaviors.From the perspective of risk-averse decision making behaviors,in order to obtain more reliable and robust decision results,the dominance relationship with risk-averse decision semantics is proposed,that is "an object is better than another object when and only when the value of an object is better than another object on all attributes".Based on this,a risk-averse multi-attribute group decision making framework for hybrid data is established.(2)The research on solution methods for attribute weight information and decision maker weight information.First,the dominance class formula is introduced based on dominance relationships.Then,under the same decision matrix,the mutual information between attributes and attribute sets is obtained according to the common information between the complement set of attribute dominance classes and the complement set of attribute set dominance classes.In addition,for the problem that only considers the common information of the above dominance class complement set and ignores the common information of the dominance class itself,the importance theorem is explored for relevant illustrations.Finally,by normalizing the mutual information between each attribute and the attribute set,the information of each attribute weight is obtained,and the same attribute has different weights of its attributes in different decision matrices.At the same time,all decision matrices given by all decision makers are aggregated to form a comprehensive decision matrix,so as to solve the decision maker weight information by applying the above method for solving the attribute weight information.(3)The research on hybrid multi-attribute group decision making problem with exact numbers,interval numbers and intuitionistic fuzzy numbers.First,the hybrid risk-averse multi-attribute group decision method and the interval number TOPSIS method are proposed based on the updated MULTIMOORA.Furthermore,the proposed method and the interval number TOPSIS method are used to conduct decision analysis and comparative analysis under the background of how automobile enterprises choose production schemes,and the effectiveness of the proposed method is verified by the validity test.The hybrid multi-attribute group decision making method established in this thesis not only can provide objective weight acquisition methods via considering the risk preference of decision makers,but also can provide robust decision conclusions.The contents in this thesis further enrich the theoretical researches of multi-attribute group decision making,and provide practical guidance for addressing complex decision making problems in real world.