An Augmented Two-stage Robust Adaptive Filtering Algorithm Based on GMWM for PPP/SINS

In dynamic precise point positioning (PPP)/ strapdown inertial navigation system (SINS) integration, abnormal observations and system model errors are both of the most important factors affecting the positioning performance. Generally, adaptive robust filtering is used to lower the impact of abnormal observations and system model errors on navigation performance, which are difficult to be accurately identified and separated. Meanwhile, the navigation performance depends greatly on the filtering model which is closely related to the error model of the inertial sensors. The inertial stochastic error model is normally determined by experience which fail to reflect the real situation of the navigation system. An enhanced two-step robust adaptive filtering algorithm is proposed to address this issue. The algorithm can respectively identify and process the system model errors and the abnormal observations in two steps. Moreover, the stochastic error of inertial sensors is modeled by an art-of-state technique called the generalized method of wavelet moments (GMWM). The results of the practical car experiments indicate that the augmented algorithm can promote the positioning performance of PPP/SINS tightly-coupled system and enhance the robustness against the system model disturbance and abnormal observations.