Fuzzy system identification method for cognitive and decision processes

The paper discusses a new fuzzy oriented method for estimating the transfer function of multivariable dynamic systems using creative interpolative fuzzy amplification for poorly informed conflicting data sets. The essential information in fuzzy rule bases, by proper techniques, can be concentrated into smaller ones. The fuzzy rule interpolation method (proposed by Koczy and Hirota (1993)) offers a possibility to obtain conclusion for an observation that does not match any of the rule antecedents, therefore, even sparse rule bases can be allowed. On the basis of the Koczy-Hirota fuzzy interpolation approach we introduce an effective transfer function estimation method using this formula in the regularization of the estimated transfer function for dynamical multivariable systems obtained from noisy, uncertain input/output data