Reducing the Computational Requirements in the Mamdani-type Fuzzy Control
József Dombi
Abstract
Fuzzy-based control systems are popular in practical applications where imprecision, subjectivity, and uncertainty can arise in the data and in the evaluation process, and it needs to be addressed. One possible solution is the fuzzy approach. However, computational requirements of these models can be extremely high, mainly defuzzification part of the Mamdani-type control. Whatever great advantages the Mamdani model has, it is closer to the human way of thinking compared to the Sugeno model, which is another popular controller. In the case where a short reaction time is required, the computational needs should be reduced. Here, we propose modified Mamdani models for this purpose. This new model is much faster while all the advantageous properties of the original Mamdani model are retained, and in some ways, it is better.