Abstract: ObjectiveParametric roll poses a significant threat to both cargo and personnel safety. The International Maritime Organization (IMO) has established the second-generation intact stability criteria and provided operational guidelines for parametric rolling in actual ships. Accurate prediction of parametric rolling in irregular waves holds important guiding significance for ship operational guidance. MethodsModel tests for the standard C11 ship were conducted in the tank of the China Ship Scientific Research Center, revealing the occurrence of parametric rolling in irregular waves under sea state 7. A dataset of parametric rolling motion time histories was constructed based on the experimental data. A data-driven model based on the LSTM neural network was developed, with nonlinear features identified through input of historical periodic data for parametric rolling characteristic recognition, and the temporal relationships optimized based on the neural network architecture. ResultsThe hyperparameters of the prediction model were adjusted with the parametric rolling amplitude as the target. A comparison of the prediction results from the training and validation sets indicates that the LSTM neural network model can be used for predicting parametric rolling in irregular waves. On the test set, the minimum mean absolute error for parametric rolling was 0.128, and the minimum maximum angular error for parametric rolling was 0.62%. ConclusionThe neural network model meets the accuracy requirements for predicting parametric rolling in irregular waves and can provide technical support for the prediction and instability warning of parametric rolling in ships