Non-Linear Predictive Control using Optimisation Techniques

In this contribution a nonlinear multivariable predictive controller is proposed where the nominal model used for control law synthesis is a neural network. The technique makes use of an on-line optimisation routine which determines the future inputs that will minimise the deviations between the desired and predicted outputs. Control is implemented in a receding horizon fashion. The paper highlights the importance of selection of the network training philosophy by application of the predictive controller to a nonlinear distillation system. The enhanced performance using the neural network based control methodology is demonstrated.