Instruments and Control
VŠB-TU Ostrava, 6.5.2003
| Annotation of paper No. 27 |
|---|
| Title: | Modelling of Helium Cryostat Using Fuzzy Adaptive System |
| Author(s): | JANOVEC, Jiøí & JURA, Pavel |
| Text: | The paper deals with modelling of real linear dynamic systems – helium cryostats – using fuzzy adaptive system, implemented on a special kind of neural networks – fuzzy neural network. For adaptation of this system we use the BACK-PROPAG method, well known adaptation algorithm for artificial neural networks. This model was implemented in MatLab, and in its fuzzy toolbox. In the first part we analyse differences and similarities between fuzzy systems (FS) and artificial neural networks (ANN). We deal with a structure of fuzzy system implemented on neural network – a fuzzy neural network, and functions of each elementary layer of a feed-forward fuzzy neural network. The we briefly describe the conventional BACK-PROPAG algorithm for training feed-back ANN. In the second part we study modelling and identification of real dynamical systems using the fuzzy adaptive algorithm and create a fuzzy adaptive model of cryostat vessel. Concrete results of simulation experiments with this model are presented. |