Data Mining In The Network Based In The Instance Of The Design System

Case-based design is an application of Case-based reasoning method in the engineering design fields. It solves current problems according to reasoning by using expertise and knowledge which have been utilized by people to work out problems similar to new ones. Case-based reasoning system requires less knowledge, expertise and integration of information than Artificial Intelligence and Expert System. Because few strict theories are based by Case-based reasoning, there are some weaknesses caused by this special character, such as difficulties in obtaining cases and establishing rule to modify best similar case etc. It becomes very hard to develop a case-based design system. Moreover design personnel's more participating in the process of design leads to the results that the case-based design system lacks intelligence and automation in the true sense, and precision and quality of the design can not be guaranteed.In this thesis Data Mining Technologies are introduced into the process of Case-based design to solve the questions appearing in the case-based design system. Then the design system is put on the internet. The author analyzes the applications of Data Mining technologies in the case-based design system, and puts forward the strategy based on regression analysis and MDOD to revise the best similar case. A new effective means to obtain design cases is proposed according to networking the design system and storing users' design results into case base. The more cases the case base holds, the better the design requirements can be met.The regression equation is obtained in this thesis by analyzing the transmission parameter of a single gear pair in the two grades cylindrical gear reducer according to multivariate linear regression analysis. Each factor's importance is decided by the partial regression quadratic sum produced by regression analysis. The design personnel can get a deeper understanding of the relationships among the product parameters by analyzing whether the factors are common parameter or characteristic parameter.On the basis of the results of regression analysis, regression equation and partial regression quadratic sum, the method of MDOD is introduced into the modification of the best similar case. The modification sequence is decided by the quantity of each factors' partial regression quadratic sum. This thesis presents the flow chart and results of the modification process. By contrast with the traditional design method, the feasibility and practicability of the method in this thesis are validated.To network the Case-based design system, a combination of C/S mode and B/S mode is put forward. Not only can the clients browse the design results through the browser, but they can also deal with the design results with the application at the end of the client. In this thesis a CBD system of two grades cylindrical gear reducer has been networked.Through the client programs the client can visualize the design results and get thethree-dimensional parameterized solid model of the two grades cylindrical gear reducer.This solid model can be used to carry out engineering analysis, and engineeringdrawings can be produced automatically through the solid model.Because the traditional Hypertext Markup Language can not meet the needs of representation of engineering data, data transmission and extracting, extensible Markup Language is introduced to be combined with Html. The Html files are used by browser and the XML files by client programs. To realize the data transmission and representation in XML format, this thesis represents a XML file format covering the data of two grades cylindrical gear reducer and practically validates its good performance.