Predicting shear stress distribution on structural surfaces under internal solitary wave loading: A deep learning perspective

The density of the ocean varies unevenly along the vertical axis. In the presence of external disturbances, internal solitary waves (ISWs) are generated. The strong shear flow field induced by ISW seriously threatens the operational safety of marine structures. Therefore, it has become a hot spot to study the force law of marine structures in ISW. The existing studies are conducted when the ISW parameters are known. However, ISW is not visible in real situations, which leads to difficulties in obtaining ISW parameters. Therefore, it is of great engineering value to accomplish real-time force prediction of marine structures without knowing the ISW parameters in advance. To fill the gap, this study proposes a novel hydrodynamic prediction model with a sensor array as the sensing system and a deep learning algorithm as the decision-making system. The model successfully achieves accurate prediction of the shear stress on the cylinder in the ISW. In addition, a technique for optimizing sensor placement is proposed. This will help identify critical regions in the graphical representations to enhance exploration of flow field information. The results demonstrate that the prediction accuracy of the optimized sensor layout scheme surpasses that of randomly deployed sensors. As a result, this study will provide an important assurance for the safe operation of marine structures.