Modeling and Optimal Control of a Batch Polymerization Reactor Using a Hybrid Stacked Recurrent Neural Network Model

Abstract

This paper presents a novel nonlinear hybrid modeling approach aimed at obtaining improvements in model performance and robustness to new data in the optimal control of a batch MMA polymerization reactor. The hybrid model contains a simplified mechanistic model that does not consider the gel effect and stacked recurrent neural networks. Stacked recurrent neural networks are built to characterize the gel effect, which is one of the most difficult parts of polymerization modeling. Sparsely sampled data on polymer quality were interpolated using a cubic spline function to generate data for neural network training. Comparative studies with the use of a single neural network show that stacked networks give superior performance and improved robustness. Optimal reactor temperature control policies are then calculated using the hybrid stacked neural network model. It is shown that the optimal control strategy based on the hybrid stacked neural network model offers much more robust performance than that based on a hybrid single neural network model.

Direct and inverse neural network modeling in free radical polymerization

“…Combining a simpli¯ed mechanistic model with a neural network model to obtain a Unauthenticated Download Date | 5/10/18 8:37 AM hybrid model can improve model representation capability [6], [7]. Tian et al [6] presented the modeling of a batch solution polymerization for a methyl methacrylate reactor using a hybrid model based on stacked recurrent neural networks.…”

“…Tian et al [6] presented the modeling of a batch solution polymerization for a methyl methacrylate reactor using a hybrid model based on stacked recurrent neural networks. The hybrid model is developed in two stages.…”

“…The hybrid model of Tian et al [6] was used to predict the in°uence of di®erent temperature policies on the behavior of the batch free radical solution polymerization of methyl methacrylate in terms of conversion, average molecular weight and polydispersity.…”

“…In order to improve the model performance and robustness in the optimal control of a batch solution MMA polymerization reactor, Tian et al [6] propose a hybrid model which contains a simpli¯ed mechanistic model that does not consider the gel e®ect and stacked recurrent neural networks to characterize the gel e®ect. Mixed order locally recurrent networks are used to build long term prediction models for a continuous solution polymerization reactor [5].…”

Identification of semi-parametric hybrid process models

“…By bringing together both existing mechanistic knowledge and data gathered from the process, a hybrid model that fuses both components has been shown, in a number of applications, to be advantageous when compared with a model formulated from either limited mechanistic knowledge or one constructed solely from the process data (Psichogios, Ungar, 1992;Thompson, Kramer, 1994;Duarte et al, 2004;Oliveira, 2004). The advantages of hybrid models have motivated a number of applications, such as the modelling of batch polymerization reactors (Tian et al, 2001), fermentation processes (Wang et al, 2009;Saraceno et al, 2009) and boilers (Rusinowski, Stanek, 2009). Besides, Teixeira et al (2007) discussed the general role of hybrid modeling in the combination of systems biology and process engineering.…”