Control by stochastic adjustment

This paper is concerned with the problem of designing an ¿adaptive controller¿ for the control of a certain class of dynamic processes. In general, these processes consist of a dynamic element governed by a nonstationary, linear differential equation. It is assumed that the designer knows only the form of the differential equation and bounds on the parameter variations but not the exact equations. A further complication of the problem is that all measurements of the state variables of the process are obscured by additive noise.