Experimental assessment of speed adaptive track control of rudder-propeller-actuated ships based on model predictive control

Abstract

Model predictive control is applied for track control of rudder-propeller-actuated ships. Speed dependent Nomoto models are proposed by analysing experimental test data, and on the basis of them predictive models are developed for path following at different speeds and trajectory tracking. Parameters of the predictive models are estimated by utilizing the least square method driven by data from manoeuvring tests, such as acceleration tests, zigzag tests and track control tests. There are multiple choices for the form of the predictive model, the speed adaptive scheme for the predictive horizon and the scheme for generating the reference path and speed, and parameter variations are made for them to investigate the effect of different parameter settings. Hundreds of free running model tests are conducted in the Towing Tank for Manoeuvres in Confined Water, Antwerp, Belgium to validate the model predictive controller's performance in shallow water. For trajectory tracking, the desired speed is assigned at each waypoint, and 16 speed plans (combinations of decelerations and accelerations) are designed for testing. In addition, real cubic obstacles are installed in the tank to generate hydrodynamic interactions with the approaching ship model, and therefore the effect of the interactions is investigated by comparing the test results with and without the presence of the obstacles. Corresponding to the experimental tests, it is proved that the controller is speed adaptive, control accuracy for the path and the speed is very high with appropriate parameter settings, rudder-propeller interaction is considered in control, and ship-obstacle interaction affects negatively the control performance.