Structure identification and parameter optimization for non-linear fuzzy modeling

This work presents a method for non-linear fuzzy model identification. The main characteristic of the method is the automatic determination of the number and position of the fuzzy sets in the domain of each variable. The resultant fuzzy rule base allows model interpretation by domain experts. The main contribution of this work is a formulation that allows the optimization of output parameters by a least-squares error (LSE) minimization. A numerical solution of the LSE problem is developed based on the singular value decomposition of the regressor matrix. The whole methodology is applied to some numerical examples found in the literature.