Adaptive fuzzy clustering and fuzzy prediction models

Abstract

This paper proposes a new fuzzy clustering technique for identification of fuzzy prediction models. An existing approach to the simultaneous determination of data partition and regression equations is modified in such a way that the shapes of clusters are changed dynamically and adaptively in the clustering process. After introducing a type of membership function, a technique for the integration of fuzzy rules is discussed. As a concrete example, a fuzzy operator model to control a rotary kiln process which treats excess sludge from a municipal wastewater treatment plant is presented. >