Towards an Efficient Approach for Mamdani Interval Type-3 Fuzzy Inference Systems

This paper is part of the increasing interest regarding the application of interval type-3 fuzzy logic in real-world problems, where a better handling of uncertainty can be useful in achieving enhanced results. The main contribution of this paper is the proposal of new methods, such as Interval Type-3 Reduction and a practical way for modeling Interval Type-3 Membership Functions, based on the Footprint of Uncertainty (FOU) and Core of Uncertainty (COU) concepts, which reduce the gap between the theory and the practical implementation of Mamdani Interval Type-3 Fuzzy Systems. The main aim of the paper is not proving the superiority of Interval Type-3 Fuzzy Systems but providing a framework and a comprehensive illustration of the theory concepts to help future research work in developing optimization methodologies and new applications for this kind of systems, as well as finding their potential applicability, which can result from their ability in handling more complex uncertainty. Simulation results with two illustrative application examples show the potential of the presented approach in achieving an efficient implementation of Interval type-3 fuzzy systems.