A PARALLEL FIXED-POINT PREDICTIVE CONTROLLER

A parallel fixed-point implementation of a general predictive controll er is derived. It is based on the concepts of predictive control outli ned by Chisci, Zappa and Mosca. These controllers use parallel identif ication arrays as the basic building blocks. In this paper the largest part of the standard information filter array is replaced with a simi lar array which uses normalized data suitable for fixed-point VLSI imp lementation. The normalization ensures that all the data are within th e range [ - 1, 1]. The controller includes an effective regularization scheme which allows a priori information about the plant and the cont roller to be included and kept permanently present in the identificati on scheme. Users can select the level of the influence of this a prior i information on the controller adaptation relative to the effect of t he measured data. The resultant architecture is suitable for fixed-poi nt VLSI or DSP application aiming for low power and minimal complexity or maximal speed, (C) 1997 by John Wiley & Sons, Ltd.