Frequency divided adaptive Kalman filter in guidance law recognition
Runle Du, Liu Jia-qi, Zhifeng Li, Niu Zhen-hong · 2016
Recognition of guidance law on non-cooperative vehicle is drawing research interest among the literature, while insufficient accuracy of modeling and measurement makes it a hard problem. In this paper, a Frequency Divided Adaptive Kalman Filter is proposed to implement guidance law recognition. In FD-AKF, a Low Pass Filter is applied to extract the noise directly from state vector and measurement vector, and estimate the co-variance matrices of observation noise and prediction noise online. Without intertwined iterating equation of prediction noise and observation noise, FD-AKF decouples the estimation on prediction noise and observation noise, and thus avoids the risk of bias and divergence. Simulation results show that, FD-AKF improves robustness of guidance recognition significantly over Kalman Filter and Sage-Husa.