On-line adaptive neural networks for ship motion control

An online neural-net control system, in which learning and control are independently carried out, is proposed for the problem of ship motion control, including roll, yaw and sway stabilization at the same time. Disturbance models, including roll moment, yaw moment and sway force induced by sea wave and wind, are presented by the experimental data in tank. With the three disturbance models as inputs, a recurrent neural network is proposed to approach the forward model of the real ship, and the real time recurrent learning algorithm is described to train the forward model. Then neural-net controller is presented to reduce the roll, yaw and sway synthetically. This paper proposes the adaptation process of control system and applies it to the ship HD702. The approaching accuracy of forward model network and the synthetic control effect of the three motions are investigated.