Underactuated ship way-points track control using repetitive learning neurofuzzy

In order to dear with the large inertia and slow responsiveness to rudder changes of the underactuated ship, an on-line trained repetitive learning control scheme, which can be used to control a class of nonlinear ship movement systems, is proposed for ship way-points tracking control. A neurofuzzy learning algorithm is developed for track-keeping and track-changing. The controller uses fuzzy inference system to mimick experienced human operator and the back-propagation gradient descent method to update the network parameters through ship running. The convergence of the new method was mathematically proved. The design is independent of ship mathematical mode and enables real time control of a underactuated ship under disturbances. Applications are demonstrated and the approaches is verified efficient.