A soft computing approach to the intelligent control
Vitoantonio Bevilacqua, Emanuele Grasso, Giuseppe Mastronardi, Leonardo Riccardi · 2006
This paper shows how "soft computing" could be useful to solve control problems. For this aim the dynamic of temperature in a room has been modelled, simulating it by the use of an artificial neural network (ANN) opportunely trained. Then, using this model, two main kinds of controllers have been tuned using a genetic algorithm: a standard PID and a fuzzy PID. Then the advantage of fuzzy systems, intended as "supervisors" to standard PID controllers, was experimented. A final comparison shows that fuzzy systems, if well tuned, could give great results in control.