Treatment of bias in suboptimal filter analysis and recursive filtering
T. E. SHIRLEY · International Journal of Control · 1978
Suboptimal filter covariance analysis for problems containing a large number of bias parameters is considered. Using a decomposition property of the truth model covariance matrix, it is shown that the effects of unmodelled bias parameters upon the non-bias parameters can be easily studied, provided the control is impulsive. The results provide a more efficient formulation of the covariance analysis problem and are generalizations of similar results for the optimal estimation problem. An alternative proof of bias separation in the Kalman filter is also given.