Identification of nonlinear systems using regular fuzzy c-elliptotype clustering

In many fuzzy control applications regular fuzzy systems are used, i.e. the Takagi-Sugeno rule bases with equidistant membership functions of equal shape. For the direct extraction of regular fuzzy systems from measured data a regular fuzzy c-elliptotype clustering algorithm is developed. In contrast to the conventional fuzzy c-elliptotype clustering, the modified algorithm identifies clusters located on a regular grid. Regular fuzzy clustering has a low computational complexity and good convergence properties.