Adaptive tracking algorithm based on maneuver detection for multi-stage ballistic target boost phase

The filtering error will jump largely in strong maneuvering periods for multi-stage ballistic target boost phase tracking using unscented Kalman filter(UKF) based on current statistic(CS) model. Theory and simulation analyses pointed out that the key reason was the filter parameters can-t be adjusted with the target maneuver strength. A new maneuver detection test statistic based on delay correlation of filtering residuals-mean was proposed, and its probability distribution was analyzed. Simulation results show that it efficiently improves the maneuver detection performance than traditional methods. Then an adaptive tracking algorithm adjusting the CS model maneuvering frequency in real time was established. The simulation results illustrate that the new algorithm can effectively restrain the error jump, speed up the filtering convergence and improve the tracking accuracy by more than 100 percent.