Adaptive Unscented Kalman Filter Based Estimation and Filtering for Dynamic Positioning with Model Uncertainties

A novel adaptive unscented Kalman filter (AUKF) is presented and applied to ship dynamic positioning (DP) system with model uncertainties of time-varying noise statistics, model mismatch and slow varying drift forces. The adaptive algorithm is proposed to simultaneously online adapt the process and measurement noise covariance by adopting the main principle of covariance matching. The measurement noise covariance is adapted based on residual covariance matching method, and then the process noise covariance is adjusted by using adaptive scaling factor. Simulation comparisons among the proposed RQAUKF, the strong tracking UKF (RSTAUKF) and the standard UKF show that the proposed RQAUKF can effectively improve the estimation accuracy and stability, and can assist the controller to obtain better control performance.