Hybrid-fuzzy modeling and identification

Abstract

In this paper a class of hybrid-fuzzy models is presented, where binary membership functions are used tocapture the hybrid behavior. We describe a hybrid-fuzzy identification methodology for non-linear hybridsystems with mixed continuous and discrete states that uses fuzzy clustering and principal componentanalysis. The method first determines the hybrid characteristic of the system inspired by an inverse formof the merge method for clusters, which makes it possible to identify the unknown switching points of aprocess based on just input–output (I/O) data. Next, using the detected switching points, a hard partitionof the I/O space is obtained. Finally, TS fuzzy models are identified as submodels for each partition. Twoillustrative examples, a hybrid-tank system and a traffic model for highways, are presented to show thebenefits of the proposed approach.