Evolutionary Fuzzy Logic Controller Schemes Suitable for the Simple Automatic Design

This paper proposes three styles of the evolutionary fuzzy logic controller (EFLC) scheme, which are adapted from the genetic-based fuzzy logic controller (GFLC) proposed by Uddin et al. (2005). The main idea of EFLC is the simple 4th order polynomial equation usages for input-output relationship calculations instead of fixed-membership function calculation style in the original GFLC. Although the usage of polynomial equations makes the proposed EFLC to have more parameters than the GFLC, the EFLC has more flexible adaptation/tuning capability and can be adapted to other applications, such as the fuzzy modeling problem. In addition, the proposed scheme still be suitable for simple automatic parameter tuning techniques, such as (1+1)-ES, and have low computation burden. The performance of the proposed scheme comparing with the GFLC is shown by two different experiments: the automatic controllers tuning and the fuzzy modeling of three-dimensional surfaces. The results show that the EFLC can be working well as the controller, and also can be applied to the fuzzy modeling problem