Abstract
Applications of neural networks in chemical process modelling and model predictive control (MPC) have been investigated for SISO systems. A multivariable, neural network modelling and MPC technique is investigated in this paper for application to a laboratory-scale chemical reactor. The reactor exhibits characteristics typical of many industrial processes, due to its nonlinearity, coupling effects among the controlled variables (temperature, pH and dissolved oxygen) and a long time-delay in the heat exchanger. Three neural models are developed for the three MISO subsystems of the process used in simulation to initially determine the control parameters and subsequently used online for the MPC of the process. Online control results are presented to illustrate the closed-loop performance of the MPC scheme. (4 pages)
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
