A Robust Adaptive Filtering Algorithm for PPP/INS Tightly Coupled Integration with Consideration of Satellite Number Variations

The Precise Point Positioning/Inertial Navigation System (PPP/INS) tightly coupled system, as a high-precision positioning method, is widely used in urban environments. In response to the problems in traditional methods, such as fixed thresholds that neglect the interference of new observations on the covariance, and the failure of the satellite number variation mechanism to suppress fluctuations in observation quality, which lead to increased positioning errors, this paper proposes a robust adaptive Kalman filter. First, abnormal observations are suppressed based on a prediction residual threshold. When the residual does not exceed the threshold, satellite number variation detection is performed, and the state noise covariance of the new observations is dynamically adjusted, thus avoiding filter misalignment caused by an excessively large initial covariance. To verify the performance of the algorithm, a vehicle-mounted dynamic experiment was designed, and three filtering schemes were compared. The experiment used real measured data from urban roads, and the results show that, compared to traditional filtering, the two-stage robust adaptive filtering algorithm significantly improves positioning accuracy. Furthermore, in scenarios with frequent satellite number changes, the two-stage robust adaptive filtering demonstrated stronger stability and adaptability.