Ship Motion Prediction Based On Optimal Neural Network Model

In recent years,with the development of the shipping industry and modern technology,research on the motion of ships at sea has become a top priority.Intense ship vibration will seriously affect the safety of helicopter take-off and landing,cargo replenishment between ships and crew members,and in severe cases,it will cause the ship to capsize and cause immeasurable losses.Therefore,it has become an important subject to accurately and effectively predict the ship’s motion attitude.In this paper,the neural network model is used to predict the motion of the ship,and an intelligent optimization algorithm is introduced to automatically optimize the hyperparameters of the neural network.The main research contents are as follows:There are problems such as low precision in the research of ship motion attitude by traditional prediction methods.A prediction model based on the Harris Hawks Optimizer(HHO)and temporal convolutional network is designed to mainly predict the roll tilt angle,pitch tilt angle,and heave displacement.When automatically optimizing the hyperparameters of the temporal convolutional network,the HHO algorithm is used to find the most suitable hyperparameters to improve the training ability and prediction effect of the network.For some problems of traditional intelligent optimization algorithms,such as the low search accuracy of a single intelligent algorithm,it is easy to fall into local optimum and so on.Design a cross-parallel intelligent optimization algorithm framework based on the Harris Eagle algorithm and Sine Cosine Algorithm(SCA).Using the excellent development mechanism of the HHO algorithm and the fast search method of the sine-cosine optimization algorithm,combining the respective advantages of the two algorithms to improve the optimization accuracy and speed of the algorithm.Aiming at the problems of a long time and large error in real-time prediction of ship motion,a real-time prediction model of ship motion named HHO-SCA-BIGRU is designed.Compared with the BIGRU network model,other models will have the problem of long time or low accuracy,and BIGRU has a good effect in these two aspects.Therefore,the proposed cross-parallel optimization algorithm is used to automatically optimize the hyperparameters of the BIGRU network to further improve the prediction ability of the network.