Neural Network‐Based Adaptive Backstepping Control of Unmanned Tandem Rotor Helicopter Considering Rotor Mass Imbalance and Crosswind Disturbance

Anti‐disturbance performance is crucial for unmanned nonlinear systems in ensuring system stability and safe flight, which has drawn tremendous research attention over past decades. This paper investigates a neural network‐based adaptive backstepping controller for attitude of an underactuated tandem rotor helicopter, in which both mass imbalance of two rotors and crosswind load disturbance is considered. First, underactuated dynamical model and detailed wind field as well as load model are established for tandem rotor helicopter system. Then, an adaptive backstepping attitude controller is developed to regulate the motion of three angles simultaneously. The mass difference of two rotors is estimated by adaptive algorithm, while radial basis function (RBF) neural network based on gradient descent algorithm is introduced to identify crosswind load in real time. In addition, the lynchpin of gradient descent algorithm is to construct the neural network approximation error, which is indirectly obtained by establishing interaction between the fitting error and the control error. Finally, comparative simulation tests with Matlab/Simulink are conducted to confirm the efficiency and superiority of the proposed control scheme.