Adaptive control by parameter-bounding techniques

A new approach to adaptive control is described. The aims are robustness and the ability to track both smooth and abrupt time variations in the plant dynamics. The technique is to compute parameter bounds for a plant model, then perform worst-case control synthesis taking account of the uncertainty in the model. Tracking of time variation relies on adaptive adjustment of bounds on the variation of the model parameters, according to the average period of validity of the model. Two control schemes are described for a single-input, single-output plant. The first is a bound-based counterpart of minimum-variance control, employing the L_$OI rather than L ₂ norm to measure performance. The second scheme places the closed-loop poles within specified bounds. In neither scheme need certainty equivalence be adopted, since the model uncertainty is taken explicitly into account in computing the control input