Optimal control of a batch emulsion copolymerisation reactor based on recurrent neural network models

A recurrent neural network based non-linear dynamic modelling and optimal control strategy for a batch emulsion copolymerisation reactor is proposed. To avoid the excessive effort and time associated with the development of a detailed mechanistic model, recurrent neural networks are used to build an empirical model to represent the complex polymerisation process. Since a recurrent neural network is trained to minimise its long range prediction errors, it can offer accurate long-range predictions which are required in batch process optimal control where the ultimate interest lies in the final product quality. Based on the developed neural network model, the sequential quadratic programming method was used to calculate the optimal temperature profile leading to a polymer product with a desired number average molecular weight, desired copolymer composition and the highest conversion. Simulation results demonstrate that this optimal control strategy can lead to improved production.