THE DESIGN OF EXPERIMENTS, TRAINING AND IMPLEMENTATION OF NONLINEAR CONTROLLERS BASED ON NEURAL NETWORKS

In the area of nonlinear predictive control, several control schemes u sing artificial neural networks have been proposed. In this work, the issues relating to the information contents of the data used to train the neural network components of these nonlinear predictive control sc hemes are considered. This raises questions about the design of experi ments. A class of feedback-feedforward nonlinear controller based on t he model predictive structure (also known as Internal Model Control, I MC, structure) is investigated. The implementation and performance of these neural network based controllers, together with comparisons to o ther nonlinear and linear controllers, are illustrated on two nonlinea r continuous-stirred-tank-reactor simulations.