A multiestimation discrete-time adaptive control with multirate sampling and supervision of the hold gain

Abstract

A multiestimation scheme with a discrete-time matching-based adaptive controller for linear time-invariant (LTI) continuous-time plant with unknown parameters is presented. Different initializations of the estimation algorithm are considered in order to accelerate the estimation process. Multirate sampling with fast input sampling is used to stabilize the discretization zeros. Although the reference tracking can be achieved by using a discrete controller, the inter-sample behavior depends on the discretization process. By an appropriate choice of a correcting gain involved in the discretization, the inter-sample ripple may be reduced. An algorithm is presented to online select the value of the correcting gain that leads to the smallest ripple.