Topics in Nonlinear and Model-based Control of Ships

Abstract

This PhD thesis considers topics within automatic motion control of ships, which has been an active research topic since the early 20th century. Specifically, the thesis aims at designing controllers to achieve a good tracking performance by handling actuator constraints, internal uncertainties and external disturbances of the ship’s inner-loop control e.g. controlling the velocity loops in order to achieve robust manoeuvrability. The thesis proposes improvements to two existing ship models, which have been found by evaluating the steady state velocities for uniformly distributed control inputs. Through this evaluation it is shown that the original ship models give rise to physically impossible motions. It is suggested to add extra terms to the damping matrices in order to overcome the issues with the existing ship models. An overview of existing performance metrics is given. Subsequently, three novel performance metrics are suggested. These performance metrics evaluate the overall energy consumption, wear and tear of the actuators and a combination of these. The proposed performance metrics are used as a tool to compare and evaluate the performance of various controllers. In addition, the use of purely nonlinear feedback strategies and combinations of linear-nonlinear feedback strategies are investigated for pose and velocity control of ships. The nonlinear feedback terms are based on a sigmoid function which limits the effects of the error term. A modification to the nonlinear feedback terms concepts is suggested by changing them from symmetric to asymmetric nonlinear feedback terms. This results in a stepping stone to handle actuator constraints. A novel motion control method is suggested in order to handle magnitude constraints of the actuator. This is based on a simplified version of the collision avoidance algorithm called dynamic window. The dynamic window algorithm was originally developed for collision avoidance for mobile robots. This control method is suggested for 2 and 3 degrees-of-freedom motion control. Here, the benefits and limitations of both the design and results are discussed. Some state-of-the-art adaptive control algorithms have been applied to a mathematical model of a ship to see if it is possible to accommodate for internal uncertainties and external disturbances. The performance of the considered adaptive control algorithms have been checked both in a numeral simulation and experimental environment. Finally, experimental work in the Marine Cybernetics laboratory and onboard the research vessel Gunnerus is described. Here, the equipment and software of the laboratory and ship are presented and discussed. Additionally, all the experimental results from the publications in Appendix A are summarised here. The thesis is organized as a mix between a monograph and an article collection. It includes eight conference papers, two published journal paper. One additional paper is mentioned, but is outside the scope of this thesis.