인공신경망을 활용한 저해상도 메쉬 기반 고해상도 메쉬에서의 충전시간 예측 및 CAE 해석 가속화에 관한 연구

Abstract

Injection molding CAE simulations are essential for optimizing the manufacturing process, but their high computational cost remains a challenge. To address this issue, this study proposes a method to accelerate CAE analysis by predicting high-resolution Fill Time from low-resolution mesh data. Specifically, we utilize rice bowl-shaped geometries as our dataset and generate simulation data with varying mesh edge lengths. Traditional interpolation methods are commonly used to estimate high-resolution results but often fail to capture complex Fill Time distributions accurately. In contrast, we employ an Artificial Neural Network (ANN) to improve prediction accuracy by learning the relationship between different mesh resolutions. The proposed approach significantly reduces computational time while maintaining high accuracy, demonstrating its potential for efficient injection molding simulations.