Improved genetic algorithm for 2D resin flow model optimization in VARTM process

Abstract

Abstract In this study, a combination of block-centered grid modeling and an enhanced genetic algorithm (GA) is introduced with the aim of optimizing the random permeability field within the Vacuum Assisted Resin Transfer Molding (VARTM) infusion model to enhance the accuracy of predicted resin flow distribution. Within the established 2D-VARTM model, random permeability values in the x and y directions are assigned to each grid. The model is then solved using the central difference method in conjunction with the upstream weighting method to predict the resin flow distribution. Subsequently, an improved GA based on heuristic mutation strategies was designed and validated. This algorithm employs the discrepancy between model predictions and actual sampling results as its fitness function and integrates heuristic strategies for iterative optimization. Simulation results revealed a significant improvement in the predictive accuracy of the model, with a jump from an initial 87.49%–97.19%. In practical applications, the predictive accuracy of the model reached 95.25%. This research offers an effective optimization approach for VARTM models and underscores the potential applicability of the enhanced GA in related fields.