Neural network predictive model based NMPC for ship path following considering amplitude and rate constraints

Abstract

In many applications, it is of primary importance to drive an object along an expected path. For different controlled plant, the applied path following methods are different. Chasing-point is a common low-speed ship path following operation, especially when the path is not straight. Many conventional and adaptive control methods and schemes are presented for the ship path following control system, which are mostly unconstrained control scheme. But usually the ship motion control systems are required to consider system constraints. As classical constrained control method, nonlinear model predictive control (NMPC) is of great meaning practically and economically in ship path following field. In view of the difficulty to obtain accurate ship model, in this paper, an Elman neural network approximated system dynamics is adopted to the NMPC controller for chasing-point method based curve path following. In consideration of the ability of the NMPC scheme to handle constraints directly, the amplitude and rate constraints on input are considered to improve control performance.