AN ADAPTIVE-CONTROL FOR LINEAR UNCERTAIN MULTIINPUT-MULTIOUTPUT SYSTEMS WITH ROBUST PERFORMANCE

This paper proposes an adaptive controller design method for linear mu ltiinput-multioutput systems with unknown parameters, an accurate mode l of which is not available. The proposed adaptive control scheme resu lts in the prevention of system performance deterioration caused by th e parameter uncertainties. In the proposed method, it is not necessary to estimate unknown parameters and/or to adjust the controller parame ters. First, a candidate for an approximate model to describe the unce rtain parts of the plant is introduced and a target model which the co ntrolled system should track is constructed based on the known nominal system and this approximate signal. Next, this tracking problem is ch anged to the linear quadratic (LQ) control one for the augmented syste m. This target trajectory is determined automatically to be close to t he optimal response for the nominal system. It is shown that the asymp totic stability of the resulting total system is ensured if certain st ability conditions are satisfied. In addition, an observer-based adapt ive control method and the stability of the combined plant-observer-co ntroller system also are discussed when not all the states can be meas ured.