Linear state estimation with parameter uncertainties
Karuppanan Srinivasan, William J. Vetter · International Journal of Systems Science · 1973
The filtering equations, i.e. the recursive relations governing the computation of the conditional mean and covariance of the state variables, when the parameters in the state and observation model are uncertain or random, are of interest in determining the sensitivity of the state estimates to parameter uncertainties. In this paper a set of filtering equations are derived for the conditions (i) when the parameter matrices in the state and observation model are stochastic and correlated, and (ii) when the parameter matrices are stochastic and statistically independent of each other.