To improve the multi-sensor fusion performance of vessel dynamic positioning

a closed-loop hierarchical fusion algorithm is proposed.The presented fusion algorithm has local and global estimators.The Kalman filter is used in both the local and global estimates.The global fusion is fed back to the local filters.Mapping relationships between the local and global estimates are defined by environment conditions derived from each local estimator.Adjusting factors based on the local and global estimates covariances are introduced.The gain of the global fusion is adjusted optimally according to the mapping matrices and adjusting factors.The global estimate is included in local estimates.The proposed algorithm combines the local and global estimates into one closed-loop mutual compensation system.By the vessel semi-physical simulation system the validities of the proposed algorithm are verified.