A novel varible gain unscented kalman filter and its application in the integrated navigation system

摘要

The unscented kalman filter is a widely used nonlinear filter in nonlinear system. But because of inaccuracies of system modeling and other reasons, when the disturbance or observation anomaly appears, UKF filtering algorithm does not have the ability of tracking the mutation state of system, so the system is likely to become unstable. In this paper, the reasons for cause these problems of UKF are analysed firstly, and then, some improvements are made to it. Then, this paper gives a detailed introduction of unscented transform, according the theory of strong tracking filter, puts forward a new kind of variable gain unscented kalman filter. At last, this paper does some simulation experiment to compare variable gain UKF filter with standard UKF. The results show that the variable gain UKF has the ability of tracking the mutation state of system when the disturbance or observation anomaly appears, and variable gain UKF really makes the system more robust and stable.