A nonlinear adaptive minimum-variance controller with application to a turbo-generator set

A new nonlinear adaptive minimum-variance controller for stochastically disturbed nonlinear systems is proposed. The identification is based on a set of discrete-time nonlinear difference equations. The relation between input- and output-sequences is identified using a Kolmogorov-Garbor polynomial description. A direct self-selection algorithm is presented for selection of significant terms of the process model for control purposes. The optimal system representation is found automatically by sequentially isolating the most statistically significant terms and adding them to the process description in a recursive manner. The optimal nonlinear model structure and a performance index which minimizes the variance of the output signal is used to derive a control scheme. Based on this an indirect parameter adaptive controller is designed. Efficiency of the proposed adaptive control scheme is demonstrated by simulation studies and control of a turbo-generator set