Research on SINS Initial Alignment and the Methods for the Parameter γ Selection Based on H∞Filtering Technique

ZHANG Tao,XU Xiao-su
2009-01-01
Abstract:The H∞ filtering was introduced to solve the problems that KALMAN filtering accuracy is not high or even divergent when system model is uncertain or noise statistics characteristics are not accurate. The basic theory and recursive formulas of H∞ filtering were introduced and compared with KALMAN filtering in essence. The initial alignment model of 12-dimensional states was established and simulated comparing with KALMAN filtering under the conditions of white noise and color noise. Three representative parameters γ were selected for comparing simulation in order to illustrate the impact of the selection of γ at the speed and accuracy of initial alignment. The simulation results showed that in the circumstances of white noise the accuracy of two filters are almost the same but KALMAN filter is slightly better. In the circumstances of color noise the anti-interference of H∞ filter is obviously stronger than KALMAN filter. At the same time, the average performance (accuracy) and robust performance of the system are balanced by selecting the parameter γ properly.
What problem does this paper attempt to address?