| The purpose of this study is to develop an integrated simulation-based optimization procedure that can intelligently, automatically, and adaptively determine the optimal process conditions for injection molding without user intervention. After developing a three-dimensional (3D) mold filling simulation program using the finite volume method (FVM) and parallel computing, and performing some speedup tests, it became obvious that an alternative approach (other than running full-fledged simulation) was necessary in order to provide the optimal process conditions within a reasonable timeframe.; The idea proposed in this study is to use Gaussian process (GP) approach to establish a surrogate model (or surrogate models in the case of multi-objective optimization) to approximate the CPU-intensive 3D simulations so that the surrogate model can capture the characteristics of injection molding simulation with minimum computational resources, thereby allowing quick iterative evaluation and system-level optimization. Based on the Bayesian probability and inference approach, the GP surrogate model provides both predictions and an estimate of the confidence (variance) for predictions simultaneously, thus suggesting a direction as to where additional training samples could be added to further improve the surrogate model. Once the surrogate model is satisfactorily established, a hybrid genetic algorithm (GA) or a multi-objective optimization GA is used to evaluate the surrogate model(s) to search for the global optimal solutions for the single or multiple objectives in a concurrent fashion, respectively. In this work, the applicability of the proposed optimization technique is investigated and the implementation of the process optimization at the system level has been developed. The proposed procedure has been tested based on function approximations and applied to a number of injection molding process optimization applications.; With the help of this adaptive simulation-based optimization system, the overall optimization task can be accomplished quickly and intelligently. The system provides design exploration and optimization technology to ensure that an optimal solution is discovered that meets or exceeds all requirements. With this system, it can greatly reduce design cycle time and manufacturing cost, and significantly improve injection molded part performance, quality, and reliability. |
