Neural Network Based Finite-time Super Twisting Sliding Mode Control For Quad-rotor UAV With Parameter Uncertainty and Mismatched Disturbance

This paper proposed a Finite Time Super Twisting Sliding Mode Control (FTSTSMC) method for the quad-rotor UAVs (QUAVs) with disturbances (matched disturbance and mismatched disturbance), uncertainties and actuator faults. Two nonlinear observers for different disturbances are proposed, the first nonlinear observer is used to estimate mismatched disturbance, and the second one is used to jointly estimate the QUAVs states and actuator faults. Further, radial basis function neural networks are employed to estimate uncertainties in system models, with weight estimated by adaptive control law. Finally, an FTSTSMC strategy is proposed to guarantee the flight performance, while ensuring the robustness of the controller, the chattering effect can be effectively reduced. Numerical simulations show the effectiveness and desired control performance of the proposed method.