Optimal feedback characteristics from stochastic automaton models

Abstract

The computation of an optimal feedback control policy for a nonlinear system disturbed by multiplicative noise may be facilitated by the use of a stochastic automaton as a system model An algorithm is presented which makes use both of the linearity properties of the automaton and of the continuity properties of the original system model to determine an automaton feedback controller which closely approximates the optimal continuous state controller for a stationary stochastic process. Its application to a specific problem involving nonlinear conditionally stable dynamics and multiplicative noise is considered.