Soft computing based control approach applied to an under actuated system.

This paper deals with the soft computing based approach for controlling an inverted pendulum system. So, two coupled controllers are used. The first controller for the angle is based on a conventional fuzzy PID-type (FPID) and the second for the position control uses a self-tunable fuzzy inference system PD-type (STFIS PD-type). In this last controller, we propose to learn the singletons part of the fuzzy decision rules of a zero order Takagi-Sugeno FIS, by an on-line learning method supervised by an automatic approach. After the learning step, a clustering technique is used to extract the rule table in a linguistic form. The clustering result makes it possible to evaluate the consistency of the behavior of the STFIS controller. In addition to show the efficiency and the robustness of the proposed control approach, the STFIS controller is compared to an ANFIS controller which is confirmed by the simulation results.