An Improved EKF Algorithm Considering Model Errors of Linearization

Extended Kalman filtering(EKF) is an effective method for nonlinear problem.However,linearization error is introduced in the process of transforming the nonlinear problem to the linear one.For the linear model error problem,the nonlinear predictive filtering(NPF) is used to predict the linear model error in the nonlinear problem,which is caused by extended Kalman filtering(EKF).The statistic properties of predicted model errors is combined with the process noise in the standard EKF to make the model more exact.Lastly,the performance of NPEKF and EKF is compared by an simulation example.The results show the validity of the compensation algorithm of model errors.