Modeling and Prediction of Ship Trajectory in Inland River Crossing Channels Based on GRU Neural Network

With the development of inland river transportation, the navigable conditions of inland waterways are becoming more and more complicated. Therefore, navigation with the aid of navigator alone is not enough. It is also necessary to make reasonable decisions on the navigation of the ship based on the future navigation behavior of the ship. Traditional trajectory prediction methods are difficult to accurately train models with the small amount of Automatic Identification System (AIS) data. In order to solve the problem of the data and the original models, this paper uses the Hermite interpolation method to expand the data. On this basis, we propose a model ship trajectory based on Gated Recurrent Unit (GRU) neural network. In the experiment, we select the AIS data of ships on the Huai'an section of the Beijing-Hangzhou Canal. Finally, Back Propagation Neural Network (BPNN) model and Support Vector Regression (SVR) model are conducted to a comparative experiments. The results show that the GRU neural network trajectory prediction model has high prediction accuracy, feasibility and effectiveness.