Identification of Hydrodynamic Coefficients in Nonlinear Maneuvering Model for an ESSO 190000 dwt Tanker by Least-Squares Support Vector Regression With Linear System

In this article, the least-squares support vector machine (LS-SVM) with linear system of equations is used to identify the hydrodynamic derivatives of a three degree-of-freedom (3-Dofs) maneuvering model for an ESSO 190000 dwt tanker. Instead of identifying the hydrodynamic derivative independently, the highly-correlated hydrodynamic derivatives are identified together by LS-SVM and subsequently distinguished in sets of linear system of equations. The issue of parameter drift is briefly discussed and diminished with the proposed approach. The training data are sampled from the simulated zigzag maneuver. The computational procedures and set up of the support vector machine are discussed in detail. A number of predicted maneuvers are compared with the reference maneuvers. Excellent accuracy and generality of the estimated maneuvering model are achieved based on the proposed system identification technique.