基于神经网络的无人机滑模飞行控制设计

Sliding Mode Flight Control for Unmanned Aerial Vehicle Based on Neural Networks

针对无人机受扰运动,基于Backstepping方法和非线性滑模控制提出了一种鲁棒神经网络飞行控制方案。对无人机姿态角速度层的系统不确定性项,采用径向基函数神经网络并对其权值进行在线调整,从而实现对其进行逼近。将回馈递推设计方法与滑模控制方法结合起来,基于神经网络的输出为无人机设计了一种回馈递推滑模飞行控制器。所设计的飞行控制器用于无人机的姿态控制,仿真结果表明所研究的无人机鲁棒神经网络飞行控制方案是有效的。

The robust neural network flight control scheme is proposed for the perturbation movement of the unmanned aerial vehicle based on backstepping method and the nonlinear sliding mode technology.For the uncertainty term of the attitude angular velocity,the radial basis function（RBF） neural networks are used to approximate it and the weight values are adjusted on line to achieve the good approximation.The robust backstepping sliding mode flight control based on the output of the neural network is presented by combining the backstepping method with sliding mode control.Finally,the proposed robust sliding mode controller is provided to the attitude control of the unmanned aerial vehicle and the results are used to demonstrate the feasibility of the proposed robust flight control method.