Direct Multivariable Model Reference Adaptive Control

Adaptive algorithms have been developed for tuning control parameters in the presence of system uncertainties and/or on line parameter variations. In explicit or indirect adaptive control, an identifier must be implemented to give estimates of these unknown parameters; these parameters are then used for control computation. In implicit or direct adaptive control,the control gains are computed without need for the intermediate identification procedure. One particular procedure that is especially attractive for multiple input multiple output systems is model reference adaptive control (MRAC). However in order to guarantee perfect tracking, certain positivity conditions need to be satisfied. Because direct MRAC procedures are easily implementable, various procedures have been developed for alleviating the satisfaction of these positivity conditions. Such procedures will be presented and illustrated with examples that apply MRAC to drug infusion control, robotics, and large structural systems.