Adaptive Filtering Algorithms of Dynamic Model

  Objectives  Kalman filter is frequently used in global navigation satellite system kinematic positioning applications. However, due to the irregularity of carriers' movement, the dynamic model is often deviated and the positioning accuracy is decreased.

  Methods  To solve this problem, two adaptive filtering algorithms are proposed to weaken the effects of dynamic model bias based on the estimated turning rate of coordinated turn (CT) model. One is the filtering algorithm that combines CT model with an improved ellipsoid constraint equation. The other is a 3D turning model for real-time estimation of the turning rate through the analysis of the carriers' movement. An adaptive filtering algorithm that combines 3D turning model and the adaptive factor constructed by the innovation vector is proposed.

  Results  The experimental results illustrate that the two algorithms can control the influence of the dynamic model errors well under different maneuvering conditions, and their accuracy is significantly better than that of standard Kalman filter and the filtering algorithm combining CT model with constant velocity model.

  Conclusions  In particular, the second algorithm not only improves the accuracy of dynamic model by adaptive estimation, but also further controls the disturbance influence of dynamic model by adaptive factor, which significantly enhances the accuracy and reliability of the navigation solutions.