Discrete Deadbeat Control of a Plant Pressure Through Identification by Neural Networks

This paper presents an application in the area of automatic control which employs artificial neural networks techniques for identification of model plant. The process to control is the pressure of a tank in the ethanol plant. The type of control is based in deadbeat algorithm with previous calculus of coefficients of the neural networks off line. The weights were calculated for backpropagation net, the proceedings for the net and control algorithm with the real time results are explained; the weights of the net are calculated with Matlab and the control of the process is written in C++. This technique of control using neural networks resolved the problems of stability and speed.