Application of adaptive model-predictive control to a batch MMA polymerization reactor

Abstract An adaptive model-predictive control (AMPC) algorithm was applied to the temperature control of a batch polymerization reactor for polymethylmethacrylate (PMMA) and the tracking performance for set point changes was investigated. An experimental control system for the polymerization reactor was constructed with cascade structure. The process parameters of auto-regressive moving average exogenous (ARMAX) model were estimated by the recursive least squares (RLS) method with self-varying forgetting factor. To make the estimation robust, we introduced a dead-zone and normalization algorithm in the RLS estimator. The model-predictive controller of unified approach was then implemented to control the reactor temperature on the basis of this process model. This indirect adaptive model-predictive controller showed better performance than the conventional PID controller for tracking set point changes, especially in the latter part of the reaction course when the gel effect became significant. Also, the control experiments were successfully conducted for tracking the optimal temperature trajectory which would yield polymer product having desired properties.