Domain decomposition for the fuzzy modeling of dynamic systems with periodic forced oscillations

This paper extends a product-space clustering algorithm for the fuzzy modeling of a chaotic regime of the periodically perturbed pendulum. Domain decomposition of the state space reduces the computational complexity of the algorithm and creates a temporal ordering of fuzzy rules. A functional decomposition of these rules becomes the model of an indirect adaptive fuzzy controller. The controller successfully trains the chaotic pendulum to follow a periodic reference trajectory.