http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
대면적 유체 코팅을 위한 기계학습 기반의 블레이드 코팅 조건 최적화
송륜근(Ryungeun Song),어솔(Sole Eo),이진기(Jinkee Lee) 대한기계학회 2021 대한기계학회 춘추학술대회 Vol.2021 No.11
Large-scale liquid coating is an important technique in various industrial fields such as the fabricating of functional panels or surfaces. Blade coating with continuous liquid supply is one of the most cost-effective methods for large-scale coating. In order to coat liquid without defects, it is essential to stably maintain the coating bead trapped under the blade, but numerous experiments are required to obtain coating conditions that guide the appropriate operating parameters. To relieve the burden on this laborious work, we present a novel strategy to acquire coating conditions via physics-informed neural networks (PINNs). Whereas standard neural networks (NNs) predict the coating performance directly from the operating parameters, the PINNs predict parameters related to the state of the coating bead to enhance the coating performance. When the data collected from the experiments under the various parameters were trained on both networks, it was revealed that the PINNs predict with a superior performance than the purely data-driven NNs. Finally, a parametric study was performed with a well-trained PINN-based model to present the optimal coating conditional zones, and we experimentally demonstrated that the stable coating was achieved by using the operating parameters from the found optimal coating conditional zones.