Computational Complexity of Fuzzy Controllers and Its Application to a Real-Time Loop Controller

Abstract In this paper the computational complexity of fuzzy controls is investigated. By analyzing fuzzy control programs, the computational requirements are computed for a few methods of fuzzy controls. The computational requirements are classified in terms of basic operations to be executed. From these results, it is shown that general fuzzy controls are hard to be implemented directly in a loop controller because of the slow inference speed and a large number of parameters. Instead, a simple fuzzy control module with a predetermined control table and mapping parameters is suggested. By real experiments it is shown that the suggested fuzzy control module is implementable in a loop controller, and by computer simulations it is also shown that the suggested fuzzy control module provides better performance than a PID controller.