A stern wake prediction method based on machine learning

Abstract

In designing the propulsion performance of a ship, the hull shape is designed to satisfy the required speed and vibration requirements, etc. In recent years, CFD has been used for designing, but is slow. It is important for designers to understand the relationship between hull shape and flow field. In this study, we develop a high-speed stern flow field estimation method using machine learning as a support tool to understand the relationship between hull shape and flow field. Using a pair of the hull shape and its calculation result by CFD as training data, we construct a machine learning model that predicts the stern flow field. And high-speed optimization of hull shape to fit the desired flow field using gradient method is also considered. Its validity was confirmed by several calculation results.