Abstract:

The adaptive fading Kalman filter adopts a fading factor to restrain the memory length of the Kalman filter, which can effectively restrain divergence of filtering when the system model and noise model are established inaccurately. But the existing formulas of calculating fading factors are complex, and the solving process is complicated, which is unfavourable for integrated navigation and some real time applications. In order to solve this problem, a new method of calculating fading factors based on innovation covariance is presented, which compensates the effect of inaccuracy information by rescaling of the error covariance through the fading factor. The proposed method has the little computation burden and improves the reliability of the filter arithmetic. Finally, the simulation results show the effectiveness of the new method.