抄録
In this study, the mechanization and automation of ship hull surface shaping were aimed by integrating finite element method (FEM) with artificial intelligence (AI) technologies. In the bending process of thick plates, such as those used in shipbuilding, localized heating and cooling with gas burners and cooling water are employed to form arbitrary curved steel plates. Traditionally, the process of shaping complex ship hull surfaces, especially at the bow and stern, relied on skilled workers. A system was developed that combines FEM analysis and AI technologies, particularly Bayesian optimization, to accurately simulate thermal deformation and mechanically calculate the optimal heating plan. Experimental tests were conducted using this system along with a specially developed small robot. However, a difference was identified between the analytical space and the actual space, and despite being able to create the target shape in the analytical space, various factors in the actual space prevented the desired shape from being achieved by following the calculated heating plan. Therefore, methods to reduce the gap between analytical and actual spaces were introduced in this study, allowing surfaces that closely match the target shapes to be achieved in shipyard testing.
