| dc.description.abstract | A dynamic positioning (DP) system for a marine surface vessel may be thought of as a thruster assisted parking brake, which allows the vessel to keep fixed at desired location and heading. Suchlike systems thus incorporate control algorithms, which accommodate to changing vessel dynamics, shifting environmental conditions and various operational tasks. This thesis presents mathematical models required to simulate a dynamic positioned vessel in the fish farming industry. The model was adapted to the work boat MS Skatten III, and methods for simulating environmental loads were derived. The overall goals have been to analyze a low-cost DP system, developed by SINTEF ICT, and suggest system improvements. Thus, the simulator software was designed to interface various control approaches. Supporting control methods such as wind feedforward control, bias compensation and negative acceleration feedback have been proposed to assist the original SINTEF system. In addition, an optimal LQ controller has been derived. The performances of the different control approaches are illustrated through simulation results from four different case scenarios, which are chosen deliberately to accentuate weaknesses and strengths of the various controllers. In the original SINTEF system, simple low-pass filters were used to remove measurement and process noise from the position measurements. Furthermore, vessel velocities were obtained from differentiation of these noise contaminated signals. Thus, a passive DP observer was proposed to improve wave filtering capabilities, include proper velocity estimation and supply bias state estimation. The bias state estimate was consequently used as a feedforward term in the suggested improved controllers. To enhance the setpoint change capabilities of the control system, a model-based reference feedforward controller was designed. This required a reference model system, which generated smooth position, velocity and acceleration reference trajectories. The result was a less bumpy and quicker transfer response, and the need for feedback control during the setpoint change decreased. In addition, the results from the LQ control simulations demonstrated the great benefits of having a coupled controller strategy, where the inherently couplings between the different degrees of freedom are embedded. Furthermore, a wind feedforward controller was implemented to better attenuate position perturbations caused by wind gusts. In this way, a corrective action was initiated in advance of the feedback control loop, resulting in improved phase properties for the closed-loop system and significantly improved positioning performance under presence of wind. A final contribution of this work, is the proposed negative acceleration feedback controller. By using linear acceleration measurements, better attenuation of low-frequent wave loads was obtained without sacrifying thruster usage. | nb_NO |
