The Research On Automatic Extraction Of CAM Template Based On Cluster Analysis

CAM templates play an important role in the process of NC programming, which isalso a typical application of knowledge reuse and rapid design. How to automaticallyextract a CAM template becomes a key issue that is needed to be solved in a NCprogramming solution provided for the enterprises. This thesis was based on the real casesand cluster analysis method from data mining. The main work and conclusions are asfollows:1, The concept of NC process unit was proposed when dealing with real NCprogramming cases. The NC operation is the object of study, the operating parameters areobject attributes, and the key parameters are selected as related attributes by analyzing therole of each parameter. Thus, the model of NC process unit is built, which provides database for cluster analysis.2, In the consideration that the attributes of NC process unit have interval-scaledvariables and categorical variables, the mixed attribute similarity model was used as thecomputing method. An improved similarity model based on objective weights was alsoproposed. The results of the experiments revealed that the distinguishing ability of theimproved model for different NC process units of different processes is better than that ofthe original model.3, K modes and GA-CLARANS clustering algorithms were selected and realized. Theresults of the experiments revealed that both algorithms could achieve good results.Compared with experience templates, the accuracy of k modes method was more than50%, while that of GA-CLARASN method was more than71.4%.4, A full-featured automatic extraction system of CAM templates was programmed onUG platform. The CAM templates extracted by the system for different situations ofdifferent enterprises were recognized by the engineers, which significantly enhanced thenormative of the enterprise templates and program readability, thus improved theefficiency of the NC programming.