The dual-kalman filtering and neural solutions to maneuvering estimation problems

Tracking maneuvering targets in a radar system is m ore complicated because the target accelerations cannot be directly measured. It may occur severe tracking error even diverge the estimates when the m aneuvering situations are happened. In this paper, we develop a Dual-Kalman filtering algorithm to handle the m aneuvering targets' tracking problems. In this approach, two collaborativ e Kalman filters are devised which one for pursuing the track estimation and the other for estimating the status of m aneuver. Based on this approach, the m ost approximate target's acceleration can be detected and estimated in real tim e. Moreover, it is also shown that one Com petitive Hopfield Neural Network-based data association combined with a multiple-target tracking system demonstrates target tracking capability.