Surf-riding and broaching prediction of ship sailing in regular waves by LSTM based on the data of ship motion and encounter wave

Predicting ship surf-riding, wave blocking, and broaching in oblique stern waves is essential for the deterministic motion prediction and control of marine structures. In this paper, the effectiveness of the dataset is ensured by using simulation data from the numerical method for the numerical simulation of the autopilot trimaran. The effect of the database obtained by classifying initial conditions on the prediction accuracy is studied first. Then, multi-condition learning and state vectors are used to make the neural network better predict the occurrence of surf-riding and broaching in various wave conditions. Finally, different input features are used to establish the database to train the neural network, and the influence of the database by different input features on the accuracy of the surf-riding, wave blocking, and broaching prediction is discussed and analyzed. The study ultimately finds that combining multi-condition learning with LSTM, optimizing input features, and introducing state vectors in the time series can effectively enhance the accuracy of the model's predictions.