Generating Hierarchical Fuzzy Systems

The proposed type of hierarchical fuzzy systems is based on the rule base hierarchy. A quite new algorithm for designing a hierarchical fuzzy system from a conventional fuzzy system is developed and implemented. The important part is the unification procedure. Redundant intermediate labels are found and subsequently unified. Basic requirements are fulfilled - subsystems of a hierarchical system are simple and transparent. A single high-dimensional and large rule base is replaced by a collection of low-dimensional rule bases with small number of rules. The input-output relation remains unchanged and allproperties and propositions of the original fuzzy system are preserved by the hierarchical fuzzy system.