Maneuverability indices estimation of nonlinear response model based on full-scale trial data and square root unscented Kalman filter

Abstract

Research on ship maneuvering performance provides a theoretical reference for autonomous ship navigation control. The gain and time coefficients accurately describe the dynamic performance of the ship. Given the challenge that the unscented Kalman filter cannot ensure the nonnegativity of the state covariance matrix during the filtering process, this study proposed a parameter identification scheme for a second-order nonlinear response model based on the square root unscented Kalman filter. First, the maneuverability indices of the free-running model were identified based on simulation data. Thereafter, the feasibility and convergence performance of the proposed scheme were verified. Based on full-scale trial data with uncertainty interference, the parameters of the second-order nonlinear response model were estimated for the YUKUN ship. The feasibility, effectiveness, and generalizability of the proposed scheme were further verified. The proposed scheme was analyzed based on three perspectives: parameter convergence, estimation results of the maneuverability indices, and yaw angle prediction results. The results obtained using the proposed algorithm were compared with those obtained using the extended Kalman filter. Finally, the proposed scheme can provide a reference for online parameter identification and prior knowledge for course keeping controllers and autonomous navigation control.