Hydrodynamic parameter identification for ship manoeuvring mathematical models using a Bayesian approach

Abstract

A novel identification approach to nonlinear ship manoeuvring models based on Bayes' rule is presented. The empirical Bayesian method is used to clean simulated polluted responses from a 20°/20° zigzag test. Two efficient Bayesian models, conjugate regression and semi-conjugate regression, are adopted for hydrodynamic parameter identification. To obtain other prior parameters, the Bayesian optimization (BO) algorithm is introduced in the Bayesian regression model. By using the identified model and an experimental model based on Blanke's nonlinear four-degrees-of-freedom (4-DOF) model, 10°/10° zigzag motion and 35° turning circle manoeuvring are performed. A comparison between the predicted results and the test results demonstrates that both Bayesian regression models have good generalization ability for identification, and that the conjugate Bayesian model has greater prediction accuracy.