Implementación de un sistema predictivo con redes neuronales para el control del comportamiento de la planta Festo MPS-PA

Abstract

In the present research work, the implementation of a Dynamic Matrix Predictive Controller (DMC) with Neural Networks was carried out for the level control process (liquid) of a Festo Compact Workstation plant of the hydraulics and pneumatics laboratory of the Technical University of Ambato. The dynamics of the plant was found through the training of a feedforward neural network, the training and testing data used were obtained by conducting an experiment that consists of applying different step inputs to the plant and the response of the system to said input. The algorithm implemented was that of a dynamic matrix predictive controller, for which it is necessary to know the mathematical model of the level process represented as a transfer function, said mathematical model was built using the exponential regression method by least squares.