Research On Nonlinear Kalman Filter Algorithm And Application Based On Sampling

Kalman Filter(KF) has been widely used in fields of state prediction and estimation. It is especially effective for linear system. For nonlinear system, such as aircraft attitude estimation, target tracking on reentry and battery state estimation, it needs to explore other state prediction, estimation methods. Extended Kalman Filter(EKF) makes the state prediction and estimation of nonlinear system to become a reality by using the first order linearization of nonlinear function. The principle of EKF is simple, but can only achieve first order accuracy, which may easily result in divergence for strongly nonlinear system. Another disadvantage of EKF is that it also needs to solve complicated Jacobian matrix. Nonlinear Kalman filtering method based on sampling attracts the attention of many scholars because of its higher accuracy and computational efficiency.Based on the research of Unscented Kalman Filter(UKF), Central Difference Kalman Filter(CDKF), Cubature Kalman Filter(CKF) and other sampling nonlinear Kalman filters,the main work of this dissertation consist of the following parts: Simplex Unscented Kalman Filter(SUKF) is used to estimate aircraft attitude, which saves the computing time. Then, the Simplex Unscented Quadrature Kalman Filter(SUQKF) is proposed by combining the SUKF and the Gauss-Laguerre quadrature rule and be used to ballistic target tracking on reentry.The results show that SUQKF can obtain better performance than SUKF. Finally, Simplex Unscented Quadrature Information Filter(SUQIF) is proposed by combining Extended Information Filter(EIF) and SUQKF, and air traffic control scenario is used to verify the performance of SUQIF. The results show that SUQIF has higher accuracy compared to EIF.