Relational partitioning of fuzzy rules

Abstract We discuss the process of fuzzy systems model building and describe the role played by the partitioning of the input space. We also introduce and investigate the idea of using fuzzy relations in the antecedent parts of the rules and show how to reason with these types of rules. We discuss the ability of relational antecedent type rules to provide alternative partitioning of the input space, their capacity to reduce the number of rules in the knowledge base, and their role in a natural description of input/output relations. Finally we look at the role these types of rules play in the typical fuzzy logic controller.