Adaptive Neural Network Control for 2-DOF Helicopter with Variable Learning Rate

This paper presents a novel approach to the problem of uncertain two-degree-of-freedom (2-DOF) helicopter tracking control using a radial basis function neural networks (RBFNN) control scheme with variable learning rate (VLR). The proposed VLR-RBFNN scheme is unique in that it reduces the negative impact of the traditional RBFNN on the controller, such as overshoot, wind-up, and chattering, during the dynamics uncertainties estimation. The implementation of the proposed method is a simple Butterworth filter-type function, which achieves decent control robustness and avoids a complex control design. Comparative experiments on a 2-DOF helicopter platform show the superiority and effectiveness of the proposed method.