An adaptive interacting Multiple Model filter for GNSS-based civil aviation

Modeling error of dynamic system is the primary cause for the failure of the extended Kalman filter in GNSS receiver state estimation. For civil aviation this problem becomes more critical since the aircraft dynamics varies with the flight phases. However, it can be solved in this paper using the interacting Multiple Model (IMM) algorithm. A new design of IMM EKF with behavior matched models for maneuver as well as for uniform motion during the flight is presented, including two mean-adaptive ¿current¿ statistical (CS) models with different designed parameters for high maneuver and medium maneuver respectively. Appropriate parameters for dynamic models and IMM algorithm are determined through an available flight database obtained from Engineering Technology Division of Air China Corporation. The performance of adaptive IMM EKF algorithm is evaluated using an actual flight data. In comparison with conventional methodology, the new designed IMM EKF shows significant improvements in positioning accuracy and robustness of GNSS for civil aviation.