Performance analysis of two advanced controllers for polystyrene polymerization in batch reactor

The performance of two advanced model based non-linear controllers is analyzed for the optimal setpoint tracking of free radical polymerization of styrene in batch reactors. Artificial neural network-based model predictive controller (NN-MPC) and generic model controller (GMC) are both applied for controlling the system. The recently developed hybrid model [1] as well as available literature models are utilized in the control study. The optimal minimum temperature profiles are determined based on Hamiltonian maximum principle. Different types of disturbances are artificially generated to examine the stability and robustness of the controllers. The experimental studies reveal that the performance of NN-MPC is superior over that of GMC.