| The term "adaptive control" was introduced into the control literature three decades ago, to describe a new methodology for controlling plants in the presence of structural and parametric uncertainties. Research in this area in the following years resulted in two philosophically distinct methods, referred to as direct and indirect adaptive control, for controlling linear plants with unknown parameters. While both methods have been studied extensively, very little effort has been expended towards combining them. In this thesis a synthesis of the two approaches is proposed for the first time. Based on the plant parameter estimates and the controller parameters at any instant of time, a closed-loop estimation error is defined and is in turn used in the adaptive laws for identification as well as control. It is shown that this combined approach results in an overall system, which is globally uniformly stable.; Using the new approach it is shown that the direct and indirect methods are equivalent, that is, identical results can be obtained with both methods based on the same prior information. Further, the combined approach is shown to be superior to the two component methods when used separately, due to the coupling that exists between plant parameter estimates and the values of the control parameters. In spite of the increased number of parameters which are adjusted using the new approach it is shown that parameter convergence can be achieved under the same conditions as in direct control. Finally, the problem of robust adaptive control is reexamined and it is shown that the combined approach can be used effectively in all the cases described in the literature where either direct or indirect control has been used. |
