Abstract
Abstract
An adaptive version of a novel robust predictive control for a class of non-linear systems is presented. The non-linear system is separated into linear and non-linear parts by Taylor series expansion and then the latter part is identified by a neural network, which is then compensated in the control algorithm such that feedback linearization can be achieved. Thus the influence of the non-linearity and model uncertainties may be eliminated or reduced. In the case of time-varying or unknown systems the linear part of the system model is estimated by an RLS (recursive least-squares) algorithm. Simulation results show that the proposed scheme may improve the system performance.
Get full access to this article
View all access options for this article.
