Direct Learning Fuzzy Logic Controller Using Powell’s Optimization
C. James Li, Nicolas W. Chbat, Eric Robrigado · 1995
Abstract This paper describes a fuzzy logic controller that is capable of improving its performance in the control of a nonlinear system whose dynamics are unknown or uncertain. This direct learning fuzzy controller is able to improve its performance without having to identify a model of the plant The performance of this new controller in the control of two nonlinear systems, a double pendulum and an inverted pendulum, was evaluated by simulations. In addition, the controller is implemented on a hierarchical controller platform consisting of an IBM Digital Signal Processor (DSP) and an IBM PC to control a Cartesian robot. The controller has shown fast learning and small tracking error in both experiments and simulations.