A proposal of fuzzy modeling for dynamic characteristics in state-space description

This paper considers the identification problem of the target plant in the form of state-space representation, focusing on the case that all the state variables are not accessible. We propose a class of fuzzy finite state models, whose dynamics are totally characterized by the dynamically changing set of membership grades in the fuzzy finite states. The proposed model´s identification method that iterates sub-approximator parameter identifications and fuzzy division parameter estimations is developed.