Robust H-infinity CKF/KF hybrid filtering method for SINS alignment

This study concerns the in-motion alignment in the strapdown inertial navigation system (SINS) with large misalignment angles. As the non-linear filtering method applied in the alignment model is quite computer intensive, which has a significant impact on the alignment accuracy and speed. To solve this problem, a robust H-infinity cubature Kalman filter (CKF)/KF hybrid filter (RHCHF) is proposed to lower the computational burden and strengthen the robustness. By virtue of the idea of model decomposition, the RHCHF could estimate the non-linear and linear parts of alignment model, respectively. Through the introduction of robust factor to adjust the filter parameters, it can ensure the accuracy reliably. The comparisons of the simulation and vehicle experiment demonstrate that the RHCHF could achieve the results at a significantly lower expense than the unscented Kalman filter, and obtain a high accuracy even when the statistical property of noise is uncertain or the outliers of measurement occur occasionally.