Abstract:

In response to the issue that traditional single-model filtering methods are unable to perceive the unknown acceleration that emerges during maneuver of space targets, thereby leading to a decrease in tracking accuracy, an improved two-stage unscented Kalman filter tracking approach with maneuver decision （TSUKF-MD） is proposed. This tracking method consists of two filters, namely a primary filter and an auxiliary filter. The primary filter does not take into account the unknown acceleration input of the target, while the auxiliary filter estimates the target’s acceleration input from the innovation of the primary filter. The maneuver of the target is perceived through the acceleration estimate of the auxiliary filter. When the target’s maneuver is determined, the acceleration estimate of the auxiliary filter is utilized to correct the target state estimate of the primary filter in order to deal with the target’s maneuvering actions within a specific time period. Simulation results demonstrate that this tracking method can effectively detect the target’s maneuver and estimate the target’s maneuvering acceleration, and the tracking accuracy of the target during the maneuvering segment has been significantly enhanced.

Key words: two-stage unscented Kalman filter (TSUKF), target tracking, space target, acceleration input