Diagonal Recurrent Neural Network Algorithm for Extreme Short Prediction of Ship Motion

The modeling method of the ship extreme short-term motion prediction by applying neural network theory is discussed. The diagonal recurrent neural network model and algorithm of ship motion prediction is given. With the algorithm, real ship motion prediction is simulated, and the simulation method and procedure are introduced in detail. The simulation results were compared with the real ship data and those from auto-regress method and periodic chart method, which proved the algorithm to be feasible。