Model predictive control (MPC) is control method based on the optimal problem using predictive future output values. Because using future output values, model predictive controller have ability improving quality control significantly compared to other methods. so that MPC has been studied and widely used in industry, especially in linear systems with time varying. However it is more difficult to build model predictive controller for uncertainty nonlinear systems. To applying the MPC to these systems, the research need a mechanism to update the system model. This paper deals with Takagi-Sugeno fuzzy model-based system identification techniques to build the predictive model with the advantage of being able to draw frem the observed input-output data using clustering techniques. This predictive model is not only more accurate but also faster than Madami model.