Temperature control in polystyrene polymerization reactor by using neural network model predictive algorithm

Department of Chemical Engineering, Engineering Faculty, Ankara University, 06500, Ankara, Turkey

E-mail: gozkan@eng.ankara.edu.tr

Manuscript received online 20 October 2020, revised and accepted 31 October 2020

Theoretical and experimental temperature control of styrene polymerization in a batch process is searched. In MATLAB/Simulink, heat signal is introduced to the model and reactor temperature change is recorded. An artificial neural network (NN) model was built with output changes of the disturbing process in the face of the pseudo-random binary sequence (PRBS) heat input. The neural network model predictive control algorithm is applied to maintain the desired temperature profile. Experimental control of the reactor temperature at the optimal temperature profile with the neural network model predictive algorithm was achieved successfully. Theoretical and experimental NN models utilized are compared and the representation ability of the NN models was shown.