조영인, 오민재, 석영수, 이성준, 노명일, “딥러닝을 이용한 선형 설계 초기 단계에서의 저항 추정”, 한국CDE학회 논문집, Vol. 24, No. 2, pp. 1-8, 2019.06.01
| Abstract | The initial ship design starts from a parent ship. A designer selects a suitable parent ship and modifies the hull shape for a new design. When a ship is modified, the hydrodynamic analysis such as numerical analysis, computational fluid dynamics analysis, model test, etc. is conducted, and the shape of the ship hull is changed according to the analysis results. This iteration continues until the owner’s, and Class requirements are met. During this process, it takes much time and high cost. In this paper, a deep learning model that can estimate the resistance of a ship using offset data is proposed, so that the most suitable candidate hull shape can be selected without taking much time and cost in the initial design stage. The normalization and early stopping methods are used in training to get the best training result. Also, the hyper-parameter tuning is conducted for the best accuracy of the proposed model. The accuracy of the proposed model is verified using the test data set, and it is confirmed that the resistance of a ship can be estimated using deep learning. |
|---|---|
| Publication Date | 2019-06-01 |
| Role | Coauthor |
| Category | KCI |
조영인, 오민재, 석영수, 이성준, 노명일, “딥러닝을 이용한 선형 설계 초기 단계에서의 저항 추정”, 한국CDE학회 논문집, Vol. 24, No. 2, pp. 1-8, 2019.06.01