Principal component analysis for low-dimensional reconstruction of roll-coupled sway and yaw dynamics in maneuver

Ship dynamics in maneuver is low-dimensionalized to be expressed by the linear combination of dynamic features acquired from various maneuver scenarios. To setup reference dynamic features, a converged correlation matrix of dynamics variables is obtained by the Monte Carlo simulation of random maneuver tests. By the principal component analysis, coupled dynamics between sway, roll, and yaw are identified. The physical meaning of dynamic features is identified to represent ship dynamics in steady turn, natural roll motion, and nonlinear behavior of hydrodynamic force and moment. Roll-coupled sway and yaw dynamics of systematic maneuver scenarios, i.e., turning circle, zigzag, and large angle zigzag tests, are explained with the dynamic features. In addition, the similarity of the dynamics features to the reference is suggested to find adequate maneuver scenario for data-driven modeling of ship dynamics in maneuver. •Ship dynamic features are identified from the random maneuver tests by the principal component analysis (PCA).•Ship dynamics in maneuver can be low-dimensionalized by linear combination of four principal component vectors.•The dynamic features are dominated by coupled dynamics in steady turn and natural roll motion,.•Large angle zigzag test is similar to the converged ship dynamic features with short test execution time.