Neural Adaptive Controller Applied to a VTOL Plant Using Takagi-Sugeno Fuzzy Model

In this paper, a comparison of the regularity actions between a conventional PID controller and a neuro-fuzzy PID controller, on a vertical take-off and landing (VTOL) plant, is presented. First, the VTOL model was identified using a classic step-test method. The conventional PID was designed using the controller synthesis method. Both plant and controller models were optimized using decreasing gradient technique. The neuro-fuzzy controller was developed starting from the characterization and identification of the singletons values for each gain contribution of the adaptative PID controller, which were introduced in a zero-order Takagi-Sugeno fuzzy inference system with Triangular membership functions applied to the error signal as input. Through several step-test, the stabilization time of the plant was evaluated, which was reduced in near 30 s using the neuro-fuzzy controller. Furthermore, the integral-square-error of the response plant was reduced with the fuzzy PID respect to the classic PID controller.