Dp 190 en
AI Application in PLC Algorithm
Author: Lukáš Kurka
This work brings the possibilities of implementation of artificial intelligence elements into PLC. The main goal is the implementation of neural networks, but in this work a realization of standard genetic algorithm is also shown. Neural network is shown in the function of predictor, model of a system and controller. The neural network as an inverse pendulum controller is trained in MATLAB and then using XML/XSLT transformed to structured text programming language according to IEC 61131.3 standard. Afterwards, it is implemented in Mosaic, which simulates PLC. The connection between controlled model (realized in MATLAB) and PLC is done by visualization environment Reliance.