Study On Kalman Filtering Application In Reentry Ballistic Target Tracking

Reentry ballistic trajectory tracking is the core issue of ballistic missile defense system. Tracking system performance will directly affect the success rate for ballistic missile interception. Tracking task is to restore the target movement information in the measurement signal from a mixed interference(noise) timely and accurately, which is get from the observation equipment (such as satellite, radar, sensors, etc.). This process is called filtering.The Kalman filter, which is be proposed in the early1960s, is autoregressive optimal algorithm based on state space model in the time domain. The filtering overcomes the shortcomings of the Wiener filter cannot estimate the multi-variable, non-stationary random process and can only be worked in the frequency domain. Because real-time recursion is suitable for computer calculation, it is widely used in target tracking, navigation calculations, automatic control and other fields. Kalman filter and its improved algorithm is studied in this thesis, based on reentry ballistic target tracking, which is a typical nonlinear filtering as application background.Mathematical model is the basis of the filtering problem. The model precision directly affect the filter estimation precision and stability. The thesis first starts with the kinematics model of the Kalman filter state space model, analysis of the principle and characteristics of various common target kinematics model. Combines with the characteristic of the observation equation under the sensor coordinate system, the thesis discusses the nonlinear dynamics model and the measurement model in different coordinate system transformation and studies the solutions in hybrid system for reentry ballistic target tracking problem. The state space model of reentry ballistic target is deduced in three-dimensional ENU coordinates system and two-dimensional vertical plane rectangular coordinate system, and ballistic generated simulation is carried out respectively. The thesis studies on the ballistic coefficient estimation strategy, which is an important kinetic parameters of reentry ballistic target.Kalman filters have been used to provide optimum estimates the states in the linear Gaussian dynamical systems, in recent decades how to use Kalman filter under the nonlinear condition became the hotspot and focus of the researchers concern. In the thesis, the optimal Kalman filter is deduced based on maximum a posteriori estimation criterion obtains from the Bayesian filtering, discusses the predict-revised autoregressive operation mechanism of the algorithm. As two improved Kalman filter algorithms, EKF and UKF are derived for the reentry ballistic target tracking and generates simulation respectively under the different of ballistic coefficient. In the last part, to improve the disadvantage of EKF and UKF in practical applications, a kind of bi-directional UKF filter algorithm based on Unscented Rauch-Tung-Striebe Smoother is proposed. It compares the performance of the proposed algorithm with that of the UKF using simulation. The simulation results show that URTSS-UKF outperforms UKF in terms of convergence speed and stability with acceptable higher computational load.