In the last decade new artificial intelligence methods for the modelling and control of complex systems, namely neural networks, fuzzy logic and probabilistic reasoning have drawn the interest of researchers and engineers. Recently, the advantages achievable by using combinations of these methods, which have independent origin and evolution, have been pointed out, generating a new paradigm which is now termed "soft-computing." This new methodology subsumes the capabilities of neural networks for modelling non-linear systems and for solving classification problems, the power of fuzzy-logic to...