The dynamic positioning of marine intelligent vehicles is gradually becoming the core technique to perform herculean tasks in complex sea conditions. To cope with the composite ocean disturbances and the constraint of communication resources, this paper proposes an event-triggered anti-disturbance dynamic positioning control strategy with performance optimization via reinforcement learning. Firstly, to overcome the composite disturbances caused by external environmental disturbance and internal uncertainties, the disturbance estimator with a fuzzy logic system is established. By constructing an event-triggered mechanism that relies on position and heading information, an event-triggered optimized dynamic positioning controller is designed by combining modified backstepping and reinforcement learning methods. The designed controller consists of a composite anti-disturbance controller and an optimized compensator. Thus, unnecessary communication can be avoided and the performance index is minimized simultaneously. Furthermore, the errors of the closed-loop system are rigorously proved to be bounded. At last, the validity of the proposed control scheme for a marine intelligent vehicle is illustrated by simulation and comparison studies.
Event-Triggered Optimized Anti-Disturbance Dynamic Positioning for Marine Intelligent Vehicles via Reinforcement Learning
IEEE Transactions on Intelligent Vehicles ; 9 , 12 ; 7592-7602
01.12.2024
2161110 byte
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Elsevier | 2025
|