Adaptive geometry and process optimization for injection molding using the kriging surrogate model trained by numerical simulation

An adaptive optimization method based on the kriging surrogate model has been developed to intelligently determine the optimal geometric dimensions and processing parameters for minimizing warpage in injection-molded components. The kriging surrogate model is a statistics-based interpolated technique that provides the approximate functional relationship between warpage and factors that influence warpage. In this study, it is used to be first trained by - and later replaced - the full-fledged, time-consuming numerical simulation in the optimization process. Based on this surrogate model, an adaptive iteration scheme that takes into account the predicted uncertainty is performed to improve the accuracy of the surrogate model while finding the optimum solution. The optimization process starts with a small number of initial training sample points and then adds additional key points during iterations by evaluating the correlations among the candidate points. As an example of validation and application, optimization of geometric dimensions and processing parameters for a box-shape part with different and stepwise wall thicknesses has been performed. The results demonstrate the feasibility and effectiveness of the proposed optimization method.