Optimization of Plastic Speed Meter Housing for Automobiles: Injection Molding Simulation, Taguchi Method and Machine Learning

Abstract

The plastic speed meter housing for automobiles requires accurate parts and assembly to inform the driver of their exact speed. For accurate assembly, the molded speed meter should have a minimize amount of deformation. In this study, to obtain injection molding conditions that minimize the deformation of the speed meter, the main molding conditions that cause the deformation of the speed meter were identified using the Taguchi method. By combining the confirmed molding conditions, 150 data sets were created, and machine learning was conducted using the data set. The model with the best accuracy learned through machine learning was the Linear Regression model. The results of this Linear Regression model were then validated with test data. The optimal injection molding conditions were derived by inputting 5000 molding conditions data into the learned Linear Regression model. Injection molding analysis was performed using the derived injection molding conditions, and the amount of deformation was reduced by about 6.4% compared to the case where current molding conditions were applied. The optimal molding conditions obtained by machine learning were applied to actcual molding. The amount of deformation of the mold amount of the molded speed meter housing was smaller than the amount of deformation predicted in the machine learning model.