Abstract:
A plant is considered which can change its state at random times and the observations of which are distorted by additive noise. An adaptive algorithm is determined for filtering the plant states which is optimal in a specified class and with a finite number of observations. Formulae are given for estimation of model parameters.