Application of Fuzzy Logic and Neural Networks for Modelling and Control
Wednesday 25 February 2009, 1300-1400
The application of fuzzy logic (FL) and artificial neural networks (ANNs) in modelling and control of complex, non-linear and time-varying processes satisfying the high system performance demands marks recently a considerable progress. The aim of the lecture is to present some basic theory and research results, related with the development and implementation of PI-like fuzzy logic and fuzzy neural controllers.
First, the essence of ANNs and some applications in modelling and control is discussed. Then, fuzzy sets and fuzzy logic are introduced and employed in building PI-like fuzzy logic process controller. Next, the process of anaerobic digestion of organic waste is studied. For this non-linear complex plant: 1) an ANN plant predictor is suggested; 2) a two-level fuzzy logic controller is designed, consisting of a primary fuzzy controller and a supervisory fuzzy controller for autotuning of the scaling factors; 3) a Sugeno ANN is trained on the basis of the fuzzy two-level controller in order to make a simple neural-fuzzy controller (NFC); 4) finally the Sugeno NFC is combined with the ANN plant predictor in the feedback to further improve the control system performance.
Results from simulation and real time control using MATLAB facilities are presented and the control system performance estimated and compared with the performance of designed ordinary PI controller system.