Two-step identification of non-linear manoeuvring models of marine vessels

• Published in: Ocean engineering 2012-10, Vol.53, p.72-82
• Main Authors: Revestido Herrero, Elías, Velasco González, Francisco J.
• Format: Article
• Language: English
• Subjects: Applied sciences; Exact sciences and technology; Ground, air and sea transportation, marine construction; Hypothesis testing; Marine construction; Nonlinear ship models; Prediction error methods; Regression analysis; System identification; Unscented Kalman filter
• ISSN: 0029-8018; 1873-5258
• DOI: 10.1016/j.oceaneng.2012.07.010

This paper presents the identification of non-linear ship manoeuvring models. It is a gray box approach in which some of the parameters of the models are known, and where a novel identification scheme for non-linear manoeuvring models based on two steps is proposed. In the first step, the structure of the model is selected using the stepwise method, and the parameters which present greater uncertainty are estimated. In the second step, a refinement of the estimates in the first step is carried out using a non-linear prediction error method with the unscented Kalman filter. As an application example, we consider a modern high-speed trimaran ferry using a full-scale trial and simulated data sets. ► We propose a novel identification scheme for nonlinear manoeuvring models based on two steps. ► We select the structure of the model and estimate the parameters which present greater uncertainty. ► We apply the unscented Kalman filter to estimate the model states. ► Reduction in bias and standard deviation in Monte Carlo studies show the good results of the proposed estimator. ► Model validation with real data confirms the good results obtained in the Monte Carlo studies.