Intelligent Prediction Model for Ship Roll Damping Based on CTGAN-XGBoost

In order to prevent the substantial rolling of ships in intricate and harsh marine environments, ensuring the safety of personnel and cargo during maritime transportation, this paper proposes an intelligent prediction model for ship roll damping based on CTGAN-XGBoost. Firstly, a dataset for roll damping is constructed using ship geometry model data and ship motion data to improve data quality. Then, a Conditional Table Generative Adversarial Network is used to balance and generate more roll damping data, thereby augmenting the quantity of available data. Finally, taking advantage of XGBoost's strong generalization ability and low overfitting degree, an intelligent prediction model for roll damping is established. Experimental results demonstrate that the CTGAN-XGBoost model achieves the highest accuracy with MSE and MAE values of only 0.325 and 0.451 respectively, along with an R2 value of 0.975. It significantly outperforms other comparative models in terms of predictive accuracy.