基于递归小波神经网络的UAV姿态变结构优化控制

无人机的姿态控制易受外界气流干扰和模型参数摄动影响,为了提高其姿态控制的精度和稳定度,提出了将变结构控制与递归小波神经网络相结合的优化鲁棒控制律.构建并分析了无人机的姿态运动模型,采用变结构控制来设计无人机姿态运动的稳定控制律,将递归小波神经网络加入到控制闭环回路中以实现变结构控制律的优化,减弱控制律对模型准确度的依赖性,并在仿真验证中与传统方法进行了比较.结果表明,该控制律能够提高其姿态控制的稳定性,且具有较强鲁棒性、较短收敛时间和较小能量消耗,从而证明了本文方法的有效性和可行性.

Optimized control for attitude variable structure of UAV based on recurrent wavelet neural networks

The attitude control of unmanned aerial vehicle(UAV)is susceptible to the external air flow disturbance and model parameter perturbation. In order to improve the accuracy and stability of attitude control, an optimized robust control law was proposed based on variable structure control and recurrent wavelet neural networks. The attitude motion model for UAV was constructed and analyzed. A stabilized control law for the attitude motion of UAV was designed with the variable structure control. The recurrent wavelet neural networks were added into the control closed loop. Therefore, the variable structure control law could be optimized and the dependence of control law on the model accuracy could be weakened. In addition, the comparison with the traditional methods was performed in the simulation validation. The results show that the proposed control law can improve the stability of attitude control of UAV, and has strong robustness, shorter convergence time and less energy consumption, which proves the effectiveness and feasibility of the proposed method.