A Fast Initial Alignment Method for SINS Used Adaptive Sample Size Unscented Particle Filter

A fast initial alignment method used Adaptive sample size unscented particle filter (AUPF) for SINS is proposed in this paper. As the SINS initial alignment with large misalignment angle is a typical nonlinear and non-Gaussian filtering problem, the unscented particle filter (UPF) is widely used in existing alignment technologies. UPF suffers from the inherent drawback of costly calculation, even though it is considered as the optimal nonlinear estimate method up to now. The new method overcomes the huge computational burden by embedding KLD-Sampling method into the re-sampling procedure of the standard UPF, so that the number of particles can ensure to be least dynamically. In addition, a quaternion-based nonlinear error model is established to not only describe the propagation of errors, but also play an important role in the implementation of filters. To verify the efficiency of the proposed method, the turntable test and computer simulations are conducted. The results of the simulations demonstrate that the new method can effectively reduce the calculation and improve the real-time performance.