| Abstrakt: |
The fixed maneuvering frequency and the assumed acceleration limit value in the conventional current statistical (CS) model may induce the degradation of the tracking performance of this model. In this study, a modified current statistical model adaptation filter (MCS-AF) algorithm is proposed to eliminate difficulties in accurately tracking maneuvering targets on the airport surface in a high-density, complex, and changeable airport environment. Firstly, the first-order time-dependent process model of acceleration noise is employed to calculate the real-time and online adjustable maneuvering frequency according to the first-order derivative definition. Then, the real-time and online updated acceleration variance is deduced through the kinematics theory model and position filtering residuals according to the position state estimation and acceleration change rate, thus realizing the adaptive updating of the model at the theoretical level. Finally, the model is verified based on the trajectory data of the real automatic dependent surveillance–broadcast (ADS–B) on the airport surface with unequal time intervals. The results demonstrate that the MCS model can realize adaptive parameter adjustment based on the prediction with unequal time intervals. The error accuracy of each statistical indicator of the model increases by 53.92%, 61.96%, 49.96%, and 43.22%, respectively, in dual-dimensional position tracking. Furthermore, the model was also optimized in terms of the velocity and acceleration tracking errors. |
Full text from SpringerLink