Abstract:Current methods for projectile trajectory prediction face limitations in accurately describing key physical phenomena, such as turbulence and cavitation, which occur during water entry. These limitations result in reduced prediction accuracy and applicability in complex scenarios. To address these challenges, this study proposes an efficient method for predicting projectile water-entry trajectories by integrating numerical simulation and machine learning techniques. The water-entry hydrodynamics are simulated using the Reynolds-Averaged Navier-Stokes (RANS) equations, the SST turbulence model, the Volume of Fluid (VOF) multiphase model, and the Schnerr-Sauer cavitation model. Additionally, the six-degree-of-freedom (6-DOF) rigid-body motion model and overset grid techniques are employed to analyze projectile trajectories under various water-entry angles. To improve prediction accuracy, a fully connected neural network (FCNN) is introduced as the prediction model, trained and optimized using simulation data. Experimental results show that the proposed model effectively predicts projectile trajectories under different water-entry angles, maintaining prediction errors within ±5%. Furthermore, predicted quantities such as velocity, depth, and lateral displacement exhibit high consistency with simulation data. These findings demonstrate that the proposed machine learning model significantly enhances the accuracy of water-entry trajectory predictions, offering robust engineering applications and valuable technical support for related fields.
