Temperature control of an acrylonitrile polymerization kettle using multiple models with second level adaptation. (Chinese. English summary) Zbl 1374.93191

Summary: A generalized predictive control method is developed from multiple models with second level adaptation for temperature control of the acrylonitrile polymerization process which has long time delays and large parameter uncertainties. Several adaptive models are designed for the system parameter ranges with the parameters estimated by a recursive least squares algorithm. Then, the model weights are calculated based on the parameter estimates and the prediction error of each model. Then, the parameter estimates are used as the true values of the parameters to determine the control action via the generalized predictive control algorithm. Simulation results show that this method enables a system with unknown parameters to quickly converge to the true value. The system performance and the tracking accuracy of the ideal temperature are significantly improved compared with conventional multiple model adaptive control.


93C40 Adaptive control/observation systems
92E20 Classical flows, reactions, etc. in chemistry
93C41 Control/observation systems with incomplete information
Full Text: DOI