A Nonlinear Kalman Filtering Algorithm For On-line Estimation With Reasonable Computing Burdens

In this paper, we propose a nonlinear filtering algorithm for the problem of online estimation with reasonable computing burdens. This method makes the best of the information available in process of the online estimation. Firstly, a parameter, namely estimate accuracy threshold, is defined whose value depends on the covariance matrix of the current state estimation. Then we decide which filtering method, either the more robust one, i.e., unscented Kalman filter (UKF) or the more computing efficient one, i.e., extended Kalman filter (EKF), to use for next iteration. Computer simulations are designed. The results demonstrate the efficiency of our proposed algorithm, as well as the superiority to the existing methods such as EKF and UKF for this problem.