Adaptive generalized predictive control and model reference adaptive control for CSTR Reactor

Abstract

Large numbers of industrial chemical process have nonlinear and time varying behavior, so to achieve good control properties it's necessary to use a powerful identification method that can track these variations properly. In this paper, an on- line recursive least square identification method based on ARX is used to have good knowledge about dynamic behavior of system, then for control goals two adaptive method is present: indirect adaptive control based pole placement and adaptive general predict control(GPC). The advantages of the methodologies are demonstrated on nonlinear Continuous Stirred Tank Reactor (CSTR) simulations. desired control properties is reached with a good parameter estimation and achieved results show the successful identifications and control methodologies. Result of two control strategy are compared together and advantages of Adaptive GPC is shown for time varying systems like CSTR.