Adaptive control of linear dynamic plants by a modified method of least squares (Q1095091)

The control of a discrete stochastic linear dynamic plant is considered in the case that the parameters of this plant are not known. A one- parameter family of adaptation algorithms which constitute a modification of the method of least squares is proposed for a wide rane of control strategies. The system is strongly adaptive with respect to the output and the control, i.e., the plant output variable \(y_ t\) and the control \(u_ t\) tend asymptotically in time to ideal processes \(y_{t^*}\) and \(u_{t^*}\) obtained on condition that the plant parameters are known. The conditions of strong adaptability have a fairly simple structure and they can be easily verified in applications.