Neural network based nonlinear model predictive control for ship path following

In many applications, it is of primary importance to steer an object along a desired path. For different controlled objectives and the dimension of the control forces, the path following control methods are usually classified into two kinds: the full-actuated and under-actuated control. Many onventional and adaptive control methods or schemes are presented for the path following control system of surface ships. The path following control systems in some situation are required to operate at the limits of their capabilities so as to maximize the performance. In this paper, a neural network iterative learning predictive model based nonlinear model predictive controller is designed for path following of surface ships. For a nonlinear model predictive control (NMPC) system, it can directly take the saturation constraints into account. And with the neural network iterative learning predictive model, the prediction is improved by the neural network predictive model which is learning online and is more alike the plant true model.