Double-input single-output fuzzy systems based on non-singleton fuzzifiers

It is acknowledged that if we select singleton fuzzifiers and normal fuzzy implications to construct fuzzy systems, the constructed systems would not be universal approximators. In this paper, by using non-singleton fuzzifiers, four kinds of double-input single-output (DISO) fuzzy system based on normal fuzzy implications are constructed. And it has been proved that all of the resulting systems have the property of universal approximation. Besides, the simulation results also illustrate the effective of conclusion. This study indicates that non-singleton fuzzifiers will play an important role when we construct fuzzy systems.