Abstract
For the manufacturing of thin-walled connectors, warpage represents an inherent challenge in injection molding, significantly affecting dimensional accuracy and shape consistency. This study introduces an optimization methodology that combines Latin Hypercube Sampling (LHS), numerical simulation, a DBO-BP neural network prediction model, and integrated multi-objective optimization algorithms (NSGA-II). Initially, LHS is employed to select experimental sample points, followed by numerical simulations to evaluate the influence of process parameters on the response variables. Based on the simulation outcomes and response data, a DBO-BP neural network prediction model is developed to enhance the precision of multi-objective optimization. Subsequently, the NSGA-II algorithm is utilized for multi-objective optimization to analyze the effects of various process parameter combinations on warpage, shrinkage, and clamping force, ultimately identifying the optimal Pareto front solutions. The optimization results demonstrate that the model's prediction accuracy for warpage and volume shrinkage is within 5%. The clamping force remains relatively high, with the optimal values for warpage, volume shrinkage rate, and clamping force being 0.173 mm, 7.5%, and 15.83 tons, respectively. This approach facilitates the optimization of injection molding process parameters while ensuring the quality of thin-walled connectors, thereby improving production efficiency and minimizing defects.