Identification of a nonlinear MIMO system using fuzzy and parametric models

In this paper is studied a methodology of fuzzy modeling already developed. The methodology is applied in a MIMO system of flow, level and temperature. Starting from input and output data is possible find a Takagi-Sugeno model using the algorithm fuzzy clustering Gustafson-Kessel, through of the optimal partition of the available data set in subsets with local linear behavior. The global model is generated for a set of determined rules by local linear models. The resultant fuzzy model is compared with parametric models using two performance indices.