An adaptive bilinear model-predictive control algorithm for a continuous stirred tank reactor

摘要

An adaptive bilinear model-predictive controller having a single tuning parameter, was used to control the temperature in a continuous stirred tank reactor. The parameters of the bilinear model are estimated at each sampling instant via a recursive least-square method with a variable forgetting factor. The model and the control algorithm are updated accordingly. An algorithm for the elimination of steady-state offset was also implemented. The controller performance was tested for changes in load and setpoint. The controller performance was found to be excellent when the steady-state offset compensator algorithm was implemented.