Tracking a maneuvering target with sliding acceleration meanvalue model and algorithm

Based upon the nonzero mean-value time correlation model a sliding acceleration mean- value model and algorithm using two Kalman filters in series is put forward. The first filter is designed to cope with all possible target maneuvers and gives out a sliding mean-value of acceleration. Using the sliding mean-value as the input the parameter of the second filter can be controlled adaptively to match the real states of the maneuvering target. This method widen the changable range of target maneuvering acceleration with high-precision of state estimation comparing with the ordinary nonzero mean-value time correlation model, such as "current" model. There are also no problems of the correlated measurement noise and correlated estimates of the two filters in early research work. Results from computer simulations are included to demonstrate the performance.