Study of the stability of Kalman filtering process
Yang Ya-li · Systems engineering and electronics · 2005
The stability of Kalman filter can not assure the convergence of Kalman filtering algorithm. And the traditional methods for detecting divergence of the Kalman filter also have some disadvantages, with a strict assumption that single measurement must be correct without measurement outlier. In addition, the traditional qualitative description of Kalman filter divergence is vague to some extent. Considering the probability of false measurements, a strict definition of the stability of filtering process by the mathematic limit form is first presented, according to the filtering effect. So the vagueness of the description of divergence is avoided. Finally, several detecting theorems are put forward, with fewer constraints than the traditional ways.