MODEL REFERENCE ADAPTIVE CONTROLLER FORMULATED FOR LOCAL OPTIMALITY IN NON-IDEAL USE.

A model reference adaptive controller that, if locally stable in non-ideal use, offers local optimality in its 'convergent' performance in nonideal use is considered. Our innovation is the introduction of low-pass filtering of the plant inputs and outputs to remove their spurious, uncontrollable high frequency content prior to their use in the adaptive algorithm. Global convergence in ideal use is provable. Preliminary analysis, interpretation, and simulation of reduced-order use are provided to validate the local optimality character and robustness improvement possible with this practical embellishment.