Improvement possibilities of the maximum defuzzification methods

Mamdani inference model is widely used in that kind of engineering applications, in which imprecision, subjectivity, or uncertainty should be handled in the data and in the evaluation process. Its popularity is coming from its operation is very close to the human thinking. Although, it has high computational needs. Final step of the conventional model is the defuzzification, where a crisp value is assigned to the aggregated result of the system to represent it. This value should be selected in such a way as to best represent the result of the evaluation. For this reason, to select the best fitted defuzzification method is essential. In this paper two of the commonly used defuzzification methods are introduced, and a correction value is proposed to improve the result of MOM (Mean of Maxima) defuzzification method. The proposed modification was studied both in the conventional Mamdani, and Mamdani-like model using discretized output.