Implementación de un control neuronal directo en el controlador lógico programable S7-1200 para aplicaciones industriales.

Abstract

o respond to the demands of the industry today, new advances are being made in the im plementation of unconventional control techniques (eg: adaptive control, intelligent control) seeking optimization, quality and competitiveness in production processes. The present work proposes the implementation of direct neural control in a Siemens S7-1200 programmable logic controller (PLC), an adaptive algorithm based on artificial neural networks for process control, the control algorithm will be implemented directly in the PLC and it can be used in different applications. Adaptive control is very useful for solving a variety of non-linear control problems where conventional control is not efficient, this characteristic makes these techniques attractive for a large number of control applications in industry, the algorithm is Available in the high level programming language of PLC’s SCL, this language allows flexibility for mathematical calculations. Two neuronal control algorithms were developed, one based on an ADALINE type network and the other based on a multi-layer perceptron type network (MLP), the two control al gorithms were validated in different plants together with a PID controller, where they were analyzed and The responses of each of the controls were compared.