The simple scheme of optimizing a fuzzy logic system and its application
Hyundoo Shin, Sung-O Jo · 2002
Many adaptive schemes have been developed to verify or fine-tune a given fuzzy logic system. Deviating from complex schemes such as those which optimize the parameters associated with the if-part and then-part typified by Takaki, Sugeno and Kang (1985) or which optimize the parameters associated with the if-part via genetic algorithms introduced by Karr (1991), we try a simple approach of obtaining the best width suited for if-part membership functions. We achieve this by developing an adaptive scheme. We then compare our scheme with one developed by Karr. It turns out that our scheme performs better than Karr's in approximating a rather complex system. In doing so, we also notice that our scheme takes a very little time to achieve a stated goal. This motivates us to apply our method to the control of an evolutionary system to test the possibility of using the scheme to simulate the system in real time. We demonstrate the practicality of the scheme by applying it to the control of an example given by Narendra and Parthasarathy (1990).