An Approach to Fuzzy Systems Modeling Based on Aggregation of Multi-Attribute Weighted Information

In this paper, Linguistic Models (LMs) for fuzzy systems modeling are reexamined based on aggregation of multi-attribute weighted information provided by the rules system of a fuzzy system and its Fact. First, we visit the process for constructing individual-rule based single-input single-output (SISO) algorithms in LMs. In this paper, the former class of algorithms is furnished by an alternate approach based on weighted maximum or weighted minimum aggregation of the corresponding multi-attribute weighted information instead of using the usual fuzzy-logic-based approach. For justification and for easy to mimic, we start with some discussion of various functional representations of a crisp discrete function and then verify that our approach is a suitable fuzzy interpolation of the crisp case. In this way, besides the well-known class of Mamdani-type algorithms and the class of Zadeh's logic-type algorithms in LMs are rediscovered, two classes of new type algorithms are constructed. Finally, we study the equivalence conditions between various composition based algorithms and individual-rule based algorithms.

並列關鍵字

Fuzzy systems modeling ； Linguistic Models ； Multi-attribute weighted information ； Weighted maximum ； Weighted minimum

國際替代計量

全文下載

主題瀏覽