Defuzzification: Maxima methods with improved efficiency and their performance evaluation

Fuzzy logic offers a broad range of applications in almost every field. The world is being reshaped by fuzzy systems. Defuzzification is the last phase in the fuzzy inference process, and it is responsible for converting the fuzzy output into a single numerical result. This paper proposes two new maxima methods- Best of Maxima and Proportionate Maxima. Proposed methods are evaluated on major properties- Accuracy (based on average % difference), time efficiency, continuity as well as other set of essential properties. An innovative way to analyze or test defuzzification method is introduced in which 100 setups of fuzzy sets with variations in asymmetries, amount of overlapping, values of $\alpha$-cut and type of membership functions are generated and tested. Detailed comparative results are presented which will be helpful for new researchers.