Neural network modeling of temperature behavior in an exothermic polymerization process

Abstract

Neural networks are applied to modeling the behavior of temperature caused by exothermic reactions in a polymerization process. In batch emulsion polymerization, there may occur unexpected thermal reactive runaway. Therefore, it is difficult to control the behavior of the system in order to keep uniform fine product quality in each batch job. Besides it is not easy to formulate physical expressions in the thermal transport phenomena of polymerization processes. Neural networks are used to model the energy balance in the batch polymerization process using normal operational data and additional operational data without an initiator. In this study, the input nodes of cooling and heating operations are respectively integrated to only one node to improve adaptability and efficiency of networks. It was shown that the temperature changes caused by exothermic reactions could be easily estimated and predicted by such neural networks in the complicated polymerization processes. The onset point of an exothermic reaction could be precisely distinguished. It was founded that the NN models could be applied as useful tools in developing temperature control systems for batch emulsion polymerization processes. Moreover, it was found that the number of input nodes affected the learning efficiency and the recalling adaptability in case of insufficient data in practical polymerization processes.