Ship intelligent autopilot in narrow water

An on-line trained neurofuzzy control scheme is proposed for ship autopilot in narrow water. Due to the large inertia and relatively slow responses of the ship, a single input multi-output control strategy is developed. This specialized learning neurofuzzy controller uses the back-propagation gradient descent method to update the parameters of the network through time. With a relatively modest amount of domain knowledge of the ship behaviour, the designed scheme enables real time control of a simulated nonlinear ship course-keeping and track-keeping under wind and current disturbances. The intelligent control approach is independent of the ship mathematical model.