Research Of Uncertain Decision Making Theory And Methed Based On Interval Numbers

This dissertation researches detailed relevant problems in MADM. Then, important contents of MADM: improvement on existing methods and presenting of new methods, determination on the attribute's weights based on the method of decision making , the MACD for other information representation that it can be turned into interval numbers ,and the interval intelligent decision making based on data are studied thoroughly.The innovative research findings are discussed as follows.(1) The definition of interval numbers and fuzzy numbers and the intrinsic link between them are discussed in this thesis. It proposes a new distance measure for interval numbers and the binary operations for them are given too. Comparisons of interval numbers with existing methods are presented and it reveals the intrinsic link between different methods. Then a brief introduction is also demonstrated for the interval vector on the basis of a full review the existing methods of standardization for interval decision matrix. Next it sums up the function model and relational model of multiple attribute decision making and lays the foundation for further study uncertain decision making theory and method based on interval numbers.(2) The uncertain decision making theory and method based on interval numbers are studied in the dissertation. It advances several methods such as TOPSIS, VIKOR,α-PROMETHEE for uncertain multiple attribute decision making based on interval numbers and a case study of comparing 15 bank branches in Iran was conducted to examine the applicability of these method. Then interval weight for uncertain decision making based on interval numbers are discussed and it proposes a TOPSIS methods dealing with interval weight for uncertain multiple attribute decision making based on interval numbers. And it works out a multi-attribute assignment problems in MAS to illustrate the method proposed by this article.(3) The method of determining the attribute's weights for uncertain decision making based on interval numbers are researched. Against the method of obtaining objective weight on the base of dealing directly with the original decision data currently and the method of decision making not play any role in the process, this paper preprocess firstly the original decision data with the method of decision making and determines the relative weight of attribute. According to the idea, this paper analyzes the relationship between the attribute's weights and the synthetic priority value and advances two kinds of methods for determining the attribute's weights based on the method of decision making, namely the weights determination based on extended PROMETHEE II method and projection pursuit model. This method is more suitable for the occasion of more schemes.(4) Other information representation that it can be turned into interval numbers is presented in this dissertation. It brings forward the VIKOR method based on generalized fuzzy numbers and the PROMETHEE method based on vague sets on the foundation of the interval numbers translated from generalized fuzzy numbers and vague sets.(5) The interval intelligent decision making based on data are researched. Traditional methods for decision analysis make use of the existing information to establish the function model or relation model; it is obviously not possible that the decision maker attempts to do it by establishing model with the rapid development and wide application of management information system, the increasing level of automation and the scale of production and operation for socio-economic and business, and more hard, dynamic even noise information. Therefore, this paper extends the FCM based on interval numbers for clustering algorithm, and propounds a multi-classification algorithm based on interval support vector machines. Both the knowledge-based intelligent decision analysis methods can work directly with characteristics of the space for the interval number for clustering and classification issues.