A Type 2 fuzzy system modelling algorithm

In this paper, a modified fuzzy system modelling algorithm that incorporates Type 2 fuzzy sets, which is based on intervalvalued membership degrees rather than singleton membership degrees, is proposed. The proposed algorithm is evaluated in terms of predictive performance and determination of the significance degrees and compared with other algorithms in the literature, namely Stepwise Multiple Linear Regression (SMLR) and Sugeno-Yasukawa [4] based fuzzy system modelling algorithm, i.e. Turksen-Bazoon (T-B) [3]. A nonlinear function, which is introduced as a benchmarking data set by SugenoYasukawa, is used for validating the models. The proposed algorithm outperformed the other alternatives both in terms of the root mean square error (RMSE) and in terms of the determination of the significance of the inputs. These results showed that the proposed fuzzy system modelling algorithm could effectively approximate nonlinear functions with simple fuzzy if-then rules without assuming a priori structure for the model.