| Abstrakt: |
This paper deals with adaptive fuzzy approaches for controlling a cart-pendulum system. In fact, it suggests the use of a conventional Fuzzy Proportional-Integral-Derivative (FPID) controller to control the angle of the pendulum and an unconventional fuzzy controller which is an interesting idea to control the cart position. For that, two Adaptive FPID (AFPID) controllers and a Self-Tunable Fuzzy Inference System (STFIS) controller are used. To sweep all the output space of these controllers, a set-point very rich in information is applied to cart displacement. The conclusions of the fuzzy rules of STFIS controller are on-line fitted by using an adaptive learning algorithm. The conclusions obtained are sorted and then are clustered in intervals. Simulation results show the interpretability of STFIS controller behavior in form of decision rules. Moreover, the STFIS controller is compared to the two AFPID controllers. This comparison shows that a FIS based on a best self-tuning of fuzzy rule base, i.e. STFIS controller, can replace correctly and more perfectly the role of a FPID controller with self-tuning scaling factors. So, the STFIS controller might be used to improve performances and robustness of a control system. [ABSTRACT FROM AUTHOR] |
