Research On Method Of Similarity Measure Of Mechanical Products System Based On Characteristics

In this dissertation, the problems of the similarity measure of mechanical products system are researched based on characteristics. The classification of different characteristics of many kinds of mechanical products and obtaining the parameter of different characteristics are researched. Characteristics correlation matrix is used to eliminate the strong correlation characteristics based on Pattern Recognition theory. Then the characteristics of mechanical products system are reduced again by Main Component Analytical Method. Consequently, the problem about characteristics selection of complicated mechanical products has been solved, and the calculation of similarity measure become easy.After some main methods of weight coefficient of characteristics being talked about, a new absolute objective method is established based on the key characteristics selection in Rough Set theory and fuzzy classification in Fuzzy theory. The two main methods of system similarity measure are compared by citing an instance. The relationship between similarity and difference is researched. Then, it is presented that similarity and difference of system can be used in optimization design mechanical productsThe research improves the objectivity and accuracy of the similarity measure of the mechanical products and enhances the efficiency of products design and manufacture.