Predictive control of a multi-component distillation column based on neural networks

A predictive scheme to control the top-product and bottom-product compositions of a three component distillation column is presented. A nonlinear model of the process for prediction of future outputs is implemented using a feed forward neural network (FNN). Operation of the column under different perturbations is analyzed using several alternatives of the neural-predictive-control strategy designed. The system behavior is compared with that of a PI controlled plant.