Adaptive ANN model based nonlinear control of a semi-batch polymerization reactor challenge problem

In the present work, an adaptive Artificial Neural Network (ANN) model based Generic Model Control (GMC) [ANNGMC] scheme is proposed for nonlinear processes. The proposed scheme consists of online parameter estimation of a purely data driven ANN model based on past measurements using Extended Kalman Filter, and control computation based on minimizing the deviation of predicted model output derivative from its desired reference trajectory in a GMC frame work. An industrial case study for temperature control of a multi-product semi-batch polymerization reactor posed as a challenge problem has been considered as a test bed to apply the proposed ANNGMC strategy at supervisory level, as cascade control configuration along with Proportional Integral (PI) controller i.e. ANNGMC-PI. The effect of batch to batch variations in terms of heat transfer variations, and some unmeasured disturbances i.e. impurity factor and monomer feed flow rate on control performance are evaluated for two products A and B. ANNGMC-PI control configuration has shown better controller performance in terms of temperature tracking with minimum root mean square output deviation (RMSOD) values as well as minimum normalized root mean square input deviation (NRMSID) values from the nominal input value compared to the standard PI-PI control scheme. The proposed scheme is found to be versatile although it is based on a purely data-driven model with online parameter estimation.