Research On Multiple Attribute Decision Making Method And Attributes Reduction With Interval Rough Numbers

The problem of multiple attribute decision making,which is researched mainly in the two aspects of decision-making method and attribute reduction, is a topical issue in decision making. As a new type of uncertain number,interval rough numbers can describe the uncertain information delicately and accurately. In the actual decision process, decision attribute values are sometimes given with interval rough numbers. The decision result is more reasonable and closer to the reality. Therefore, the study on the interval rough numbers multiple attribute decision making methods and attributes reduction problems has significant value in theory and practical application. From the following several aspects, the problem of decision methods and attributes reduction on interval rough numbers will be studied.(l)According to the definition and structure of interval rough numbers, a normal distribution of interval rough numbers is defined, it is possible to the general interval rough numbers transforms into a normal distribution of interval rough numbers. And the definition of normal distribution interval rough numbers possibility degree formula is given. A multiple attribute decision making method is proposed based on the possibility degree of normal distribution interval rough numbers.(2) With the idea of set pair analysis, the interval rough numbers and set pair analysis are combined, and the connection number model is constructed based on interval rough numbers. A decision-making model is proposed based on connection number, the model is simple and easy to be understanded. The specific examples are cited to illustrate the methods are reasonable and effective.(3) A interval rough numbers similarity degree with a parameter number and great similarity are defined based on interval numbers. Establishing a differentiable matrix determines whether attribute is important. then, a new reduction of attributes based on differentiable matrix is proposed. Further, the information entropy of a good nature is defined and used to measure inner significance of attributes and outer significance of attributes. A new attributes reduction is given in interval rough values information system area. Examples are given to show the effectiveness and feasibility of this approachs.