The H∞ filtering is introduced because KALMAN filtering accuracy is not high or even divergent issues when system model is uncertain or noise statistics characteristics are not accurate. H∞ filtering basic theory and recursive formulas are introduced and compared with KALMAN filtering in essence. The initial alignment model of 12-dimensional states is established and is simulated compared with KALMAN filtering under the conditions of white noise and color noise. Three representative parameters γ were selected for compared simulation in order to illustrate the impact of the selection of γ at the speed and accuracy of initial alignment. The simulation results show that in the circumstances of white noise the accuracy of two filter 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.
