International Journal of applied mathematics and computer science

online read us now

Paper details

Number 4 - December 1996
Volume 6 - 1996

Discrete-time and continuous-time generalised predictive controllers with anticipated filtration: Tuning rules

Zdzisław Kowalczuk, Piotr Suchomski, Andrzej Marcińczyk

Abstract
A general class of control algorithms based on the generalised predictive control (GPC) strategy with anticipated filtering (AF) of the control error is considered in both the discrete- and continuous-time domains. It is shown that in the discrete-time settings, under certain conditions, a solution of the AF-GPC design always exists and the design leads to stable control systems with definite closed-loop characteristics. The plant cancellation issue is taken into account. Conditions for the existence of the solution of the GPC design and the corresponding rules of tuning of the resulting controller are given. A suitable iterative procedure for a simultaneous determination of the AF-GPC design parameters (the control horizon and the order of plant cancellation, as well as the controller gain) and a root locus interpretation of the design are also supplied. The continuous-time predictive control (CGPC) has properties similar to those of the discrete-time GPC strategy. It is shown that the idea of using the anticipated filtering approach to the GPC design can also be effectively applied in the continuous-time restatement. Rules for tuning the CGPC controller, which are based on parameters of a system rate of reaction, identified by a starting phase of system step response, are given and shown to be practically effective. With the anticipated filtering, applied in both the discrete- and continuous-time frameworks, the excitation of the closed-loop system is suitably abated by performing a moderating filtration in the anticipated-time domain. The pertinence of the anticipated filtering lies in shaping the closed-loop characteristics of the control system, reducing the disagreeable control effort and, consequently, based on a certain balance obtained in the cost function, in making the λ-tuning more practicable. The proposed tuning rules are validated via simulation experiments.

Keywords
-