Recurrent Neural Network approach for data prediction of a chemical process plant

Artificial neural networks especially the Recurrent Neural Network play an important role in nonlinear system modelling and control because of their effective and efficient nonlinear mapping, nonlinear- nonparametric adaptive-learning property and parallel processing capabilities. These models can be used for data prediction, classification, clustering and adaptive control of the system. A chemical process plant is a complex and complicated system and the mathematical modelling will be a tedious process. In this work a Continuous Stirred Tank Reactor (CSTR) which is a complex non linear type of reactor in chemical industries is considered for data prediction process. In this paper an attempt is made to model the nonlinear CSTR using different ANN models which are used for data prediction. The Elman network is a good example for the recurrent neural network. A CSTR is modelled using an Elman Neural. Network and the simulations are carried out. The results imply that Elman Neural Network can be effective in data prediction with a good accuracy.