A gain adaptive fuzzy logic controller

Abstract

We propose a PI-type gain adaptive fuzzy logic controller (GAFLC). Input scaling factors (SFs) of the GA-FLC for error (e) and change of error (Δe) are updated on-line, by a single non-linear parameter computed through a unipolar sigmoid function defined on the current process states (e and Δe). The performance of the proposed GA-FLC for step set-point change and load disturbance is compared with those of a conventional PI-type FLC (FPIC) and a Ziegler-Nichols tuned (ZN-tuned) non-fuzzy PI-controller (NFPIC). Results for various higher order processes show that GA-FLC provides much improved performance over both FPIC and ZN-tuned NFPIC. © Springer-Verlag Berlin Heidelberg 2002.