A Model Reference Adaptive Control Scheme with External Controllers

Abstract

A new model reference adaptive control scheme, suitable for linear multivariable discrete time systems, is proposed. A set of parameter adaptation algorithms are developed based on the validity of perfect matching between the model and the plant, and on the availability of complete state vector. Arrangements for linear state error feedback, and time varying adaptive loop gains are made in the proposed scheme. It is shown that the scheme with such arrangements offers increased rate of decay of adaptive system errors towards the origin of the error space. A problem of on-line matrix inversion, encountered in implementing the proposed adaptation algorithms, is solved by modifying the structure of one of the adaptive controllers.