System modelling and fuzzy relational identification
José Valente de Oliveira, Joao Miranda Lemos · 2002
Fuzzy relation equations are a suitable framework for modeling physical processes. However, their applications to identification and modeling problems are still weakly explored. To fill this gap, two related critical issues are addressed: the construction of numeric/linguistic interfaces and the computation of the fuzzy relation. An optimizing algorithm is adopted for the construction of the numeric/linguistic interface. Adaptive learning is proposed for determining approximated solutions for systems of fuzzy relation equations, namely for their extended versions. Simulation results are provided showing both fast learning rates and good performance for the derived model.>