A Study On Theoretical Models And Methodologies For Multilevel Hierarchical Multicriteria Multiperson Decision Makings In Complex Fuzzy Large Systems With Their Applications

Considering fuzziness and stochastic uncertainty for complex large scale systems, this thesis investigates systematically theoretical models and methodologies for multilevel hierarchical multicriteria multiperson decision making problems, establishes a lot of original theories and methods which present effective ways to solve decision making problems for practical complex systems, and are of great value to scientific researches and real applications.Weight is an important concept and a difficult problem to solve for multicriteria decision makings. Based on the discussion of essentiality and determination for weight, two methods which are the ordered pairwise chain comparison and interval number for determining weight are proposed and established. Their principles are simple and easy to implement, and moreover, there is no need for test of consistency. Furthermore, the rank preservation and method to determine weight are explored if new objective elements are introduced.The membership function is an important concept for fuzzy sets and fuzzy multicriteria decision making problems. It is a foundation stone of fuzzy mathematics. It is an important research subject to eliminate subjectivity in the process of determination for membership grade. For this/reason, comparison analysis is made between absolute membership grade and relative membership grade. Hence, a new method to determine relative membership grade for criteria is proposed. This method is called ordered pai(?)wise chain comparison.According to the characteristics of many multiattribute decision makingproblems with time in real world, the concept of multiattribute decision makings with time series is put forward , two fuzzy models which are the linear weighted average comprehensive evaluation and comprehensive optimum seeking are established. Furthermore. comparison analysis is made between the two models.In real world, there are a lot of multiattribute decision making problems in which preference information (i. e. , weight) of the decision maker is unknown a priority. To solve these problems, the fuzzy iterative method and fuzzy neural networks comprehensive optimum seeking method for multiattribute decision making problems are established. By iterating alternately, the fuzzy iterative method can obtain both weights of attributes and optimal membership grades of alternatives. The convergence of the fuzzy iterative method is proved. Combined conventional BP learning algorithm and fuzzy optimum seeking theoretical model, the fuzzy neural networks comprehensive optimum seeking method is put forward. Contrasted with other methods, this method can obtain automatically knowledge and experience of human, and has functions of self-organization and self-learning. Furthermore, this method is easy to operate and implement.Based on the fact that there exist a lot of decision making problems which are group decision makings with multiple decision makers, we discuss multiattribute group decision making problems in which decision making matrices are unknown a priority and develop a new linear assignment method. This method requires a little information and may solve a rather large type of decision making problems.According to the need of Dalian city for water resources and economy continuative development, we establish a multiobjective decision making model for Dalian city macroeconomic water resources planning and develop a fuzzy multiobjective group decision making method which presents an effective approach to solve practical problems.Resources allocation problems are a key problem for economy management. Cooperation among decision makers in the lower level permitting, this thesis investigates this type of decentralized hierarchical resources allocation problems, explores the formation of coalitions for decision makers in the lower level and the imputation way for coalition profit, and...