Revistes Catalanes amb Accés Obert (RACO)

Refinement of a fuzzy control rule set

Antonio González Muñoz, Raúl Pérez Rodríguez


Fuzzy logic controller performance depends on the fuzzy control rule set.
This set can be obtained either by an expert or from a learning algorithm
through a set of examples. Recently, we have developed SLAVE an inductive
learning algorithm capable of identifying fuzzy systems. The refinement of
rules proposed by SLAVE (or by an expert) can be very important in order to
improve the accuracy of the model and in order to simplify the description
of the system. The refinement algorithm is based on an heuristic process of
generalization, specification, addition and elimination of rules.

Text complet: HTML