Comparative analysis of reduced Kalman filters with guaranteed estimation quality

The paper gives a comparison of two methods for synthesis of reduced Kalman filters with guaranteed estimation quality. Their distinguishing feature is the fact that the matrix calculated in the covariance channel of a synthesized suboptimal filter is an upper bound for the real covariance matrix. In the first method, guaranteed estimation is provided due to the increase in disturbing noises, and in the second, due to two factors: the increase in disturbing noises and changes in the dynamics matrix, which improves the actual accuracy of filtering in some cases.