On developing enhanced fuzzy models for nonlinear process control

A special method for system modeling is presented to develop multi-variable fuzzy models on the basis of the system's measured input and output data. The global modeling problem is proposed to be solved in two steps: by fuzzy model configuration and fuzzy model identification. The fuzzy model configuration procedure is characterized by preliminary considerations leading to the definition of the basic model structure and appropriate fuzzy operators to be applied Representing the crucial point in fuzzy modeling, the fuzzy model identification procedure is carried out by applying a special clustering method, the fuzzy c-elliptotypes method, providing the parameters of the fuzzy model. Additionally, a special rule base generation algorithm is mentioned identifying the rule base of the fuzzy model, i.e. the system-specific links between the input and the output fuzzy sets. As an application the resulting fuzzy model is finally embedded into an special conception of nonlinear fuzzy control, controlling exemplarily the output signal of an undamped one-degree-of-freedom oscillating system with nonlinear characteristics.