PREDICTIVE CONTROL OF QUALITY IN BATCH POLYMERIZATION USING HYBRID ANN MODELS

Two issues involving the methodology used for on-line control of produ ct quality in batch manufacturing processes ave addressed: the generat ion of fast, data-driven process models and the use of such process mo dels for on-line feedback control of product quality. The methodology is investigated using the example of the control of dispersity and mol ecular weight distribution in a batch reactor for emulsion polymerizat ion of vinyl acetate. An artificial neural network (ANN) is used as a model to predict the quality as a function of the manipulated variable s and on-line measurements. This model is constructed using an augment ed dataset that integrates experimental information and knowledge from a mathematical model. The proposed model is compared with other types such as a theoretical model whose key parameters are fitted to experi mental data. The hybrid ANN is superior to the parameter-fitting appro ach for this case. Experimental and simulation studies confirm the adv antage of using the proposed model and the predictive control algorith m.