Adaptive Prescribed-time Control of Dynamic Positioning Ships Based on Neural Networks

In this paper, a novel controller with prescribed-time performance is designed for dynamic positioning (DP) system of ships with model uncertainty and unknown time-varying disturbances. Initially, an error transformation function with zero initial value is introduced by constructing fixed-time funnel boundaries (FTFBs) and a fixed-time tracking performance function (FTTPF). The proposed controller ensures stable convergence of the new error, maintaining it within fixed upper and lower boundaries. When the prescribed time is reached, the system state will achieve prescribed-time (PT) stability. Secondly, by deploying radial basis function neural networks (RBF-NNs) and dynamic surface control (DSC), adaptive controller with simple forms are rationally applied to Backstepping technology, and the uncertain terms of the system are approximated online, the singularity and complexity explosion problems of the ship control system are also addressed. In addition to that, the stability analysis results of the system prove that all errors of the closed-loop system are semi-global uniformly ultimately bounded (SGUUB) stable. Finally, the simulation results on a DP ship confirm the superiority of the proposed scheme.