Sensitivity Analysis for Type-1 and Type-2 TSK Fuzzy Models

Q. Ren, L. Baron, and M. Balazinski (Canada)


Fuzzy logic system, subtractive clustering, modeling, sensitivity analysis


In this paper, subtractive clustering method is combined with least squares estimation algorithms to pre-identify a type-1 Takagi-Sugeno-Kang (TSK) fuzzy model from input/output data. Then the type-2 fuzzy theory is used to expand the type-1 model to a type-2 model. A sensitivity analysis is used to ascertain how a type-1 TSK model output depends upon the pre-initialized parameters and determine how a type-2 TSK model output depends upon spread percentages of cluster centers and consequent parameters. By using sensitivity analysis, we can check the quality of TSK models, and characterize the uncertainty associated with the TSK fuzzy models.

Important Links:

Go Back