A Design Methodology for Fuzzy System Interfaces

Abstract:

Conceptually, a fuzzy system interacting with a numerical environment has three components: a numeric/linguistic interface, a linguistic processing unit, and a linguistic/numeric interface. At these interfaces, membership functions representing linguistic terms play a top role both for the linguistic meaning provided and for the pre/post information processing introduced to the fuzzy system. Considering these issues, a set of membership function properties is postulated. Furthermore, an expert-free interface design methodology able to meet these properties, and based on the concept of optimal interfaces, is proposed. This concept simply states an equivalence between information formats (numeric and linguistic), thereby making the methodology appealing from the applicational point of view. An algorithm is developed, and brief notes on selected applications are outlined stressing relevant issues of the proposed methodology. © 1995 IEEE

Año de publicación:

1995

Keywords:

Fuente:

scopus