Stochastic optimal control with imperfectly known plant disturbances
TZYH JONG TARN · IEEE Transactions on Automatic Control · 1970
It is the purpose of this correspondence to show how filtering theory based on a Bayesian approach may be used to solve the problem of optimally controlling a linear discrete stochastic system in which the additive Gaussian plant noise has fixed but unknown variance. Selecting a reproducible type of probability density and applying dynamic programming, an exact analytical solution of the feedback control law may be found.