Ship Roll Prediction Model Based on XGBoost and its Interpretability Analysis
In order to accurately predict the ship's attitude,a nonlinear mapping relationship between various factors and ship roll motion is constructed by utilizing the eXtreme gradient boosting(XGBoost)model.Furthermore,the Shapley additive explanations(SHAP)method is employed to enhance the interpretability of the model's predicted outcomes,thereby facilitating a deeper understanding of the factors influencing ship roll motion.The results show that the XGBoost model exhibits excellent learning and predictive capabilities for the roll angle,with the coefficient of determination exceeding 0.95 for both the training and test sets.The primary factors influencing the ship's roll angle are time,speed,and damping,whereas the secondary factors include the water surface line coefficient and draft depth.The findings of this research offer valuable insights for predicting ship motion behavior and enhancing the stability of ship operations,thereby serving as a reliable reference for maritime professionals.
万方数据
维普
