Improvement of injection molding process using classification and RSM mixture model

This paper proposes the classification and RSM (Response Surface Methodology) mixture model for improving the injection molding process of smart phone camera body. To do this, we used the big data of manufacruing process condition collected from micro injection molding machines and its quality results for camera body as input and output to the tranining data set. After preprocessing step including cleaning and discretization, the feature selection was performed to select the important variables affecting the quality of a camera body in the injection molding process. At the next step, we figured out the characteristcis of variables and their relationships from classification learning models, that can be used to simplify the RSM model which has originally many variables and terms in the polynomial equations. Finally, the classification and RSM mixutre model can provide the injection molding process conditions for assuring qualified products. The proposed mixture model requires only one eighth variables and one twelveth number of experiments as compared to a traditional RSM model.