Ship Dynamics Modeling Based on Multivariate Identification by Linear Combination of Principal Components

The present study suggests a data-driven multivariate identification method based on principal component analysis and shows an application to ship dynamics modeling in maneuver. A reduced order model of ship dynamics is built by linear combination of three principal components acquired from large angle zigzag maneuver test. For a given kinematic state with three variables, a proper span is found by least square method, therefore accompanying hydrodynamic force and moment is determined. Suggested dynamics model correctly estimates hydrodynamic force and moment, thus it showed good agreement in maneuver simulation with that of conventional ship dynamics model obtained by system identification of captive model tests.