This paper propses an eclectic approach for the efficient computation of fuzzy rules based on fuzzy logic and signal processing techniques. The rules {R_r} of the MISO zero-order Takagi-Sugno fuzzy system considered, are given in the form of R_r: If X_1 is A_(r1) and ... and X_N is A_(rN) then z is c_r, where Xj are fuzzified input variables, A_(rj) are standard fuzzy sets which belong to the corresponding partition of unity {A_(ri)} and C_r is a nonfuzzy singleton term of output variable z. A relevant feature of this approach is a quantitative, signal processing based, transformation of uncertainty (imprecision) of each input Xj into an additional uncertaintly (vagueness) on the corresponding fuzzy partition {A_(rj)}. This transformation greatly simplifies the involved matching computation. Moreover, this fuzzification transformation gives a new set of linguistic terms {A_(rj)'} which is also a partition the unity.
展开▼