Continuity focused choice of maxima: Yet another defuzzification method (Q5951746)

The authors show that by introducing some minor constraints for the components of a Mamdani fuzzy controller, it is possible to define a defuzzification method that selects an element of the core of the fuzzy output set as a defuzzification value and that also guarantees the continuity of the controller. The constraint with the largest impact on the controller is the property of linguistic continuity. Although this constraint seems very natural, it is not fulfilled in most of the existing controllers. Hence, it is indicated that in general the enforcement of the property of linguistic continuity will result in an increase in the number of rules for the fuzzy controller. However, the highly efficient defuzzification method compensates for the loss in performance caused by the higher number of rules. Furthermore, the defuzzification method makes it possible to tune the controller locally. This property can enhance the overall tuning process of the fuzzy controller. As a result, the proposed controller could be a viable alternative for the fuzzy controllers existing today.