The novel impact point prediction of a ballistic target with interacting multiple models
Jae-Kyung Jung, Dong-Hwan Hwang · 2013
The threat of ballistic targets has increased rapidly in recent years. Therefore, it is essential to prepare the capabilities to predict their impact points in order to assign the firing battery to defense them effectively. Because the trajectory of a short-range ballistic target represents severe non-linear characteristics and consists of boost phase and ballistic phase, it is difficult to estimate the state and predict its impact point using single dynamic model in overlapping region. In this paper, the method to distinguish the trajectory phase from the measurement data and the method to estimate the state using a different extended Kalman filter (EKF) with interacting multiple models are proposed in order to fuse the state of a ballistic target in overlapping region. For effective the state fusion, it is necessary to merge each state from a different EKF in accordance with the mode probability depending on the residual error between the estimated state and measurement. A Monte Carlo simulation is used in the verification of the proposed method.