Performance Analysis of Stable Adaptive NARMA Controller Scheme for Furuta Pendulum

This paper presents a novel stable adaptive controller scheme for Furuta Pendulum via nonlinear auto-regressive moving-average based plant identification. During online learning for the developed controller, input-output data obtained from the rotary inverted pendulum model used to update the parameters of the NARMA controller while ensuring Schur stability for the overall closed-loop control system. The parameters of the plant model and the introduced controller are computed by minimizing the identification and output tracking errors, respectively, both of them are absolute loss functions modified with a regularization parameter. The proposed adaptive controller is tested on Furuta pendulum model and its performance is compared with the performances of proportional integral derivative controller and model reference adaptive controller.