Scholarlink

In addressing the attitude tracking control problem of the hypersonic morphing vehicle (HMV), a composite control method including feedforward and feedback control is proposed in this paper. Firstly, a feedback controller is established using the backstepping method and active disturbance rejection control (ADRC) scheme to overcome the external disturbances and model uncertainties. Next, a feedforward controller based on long short-term memory (LSTM) neural network is introduced to enhance the HMV’s response speed, disturbance rejection capability and adaptability to the aerodynamic characteristics variations and uncertainties. Then, the feedback controller is redesigned to compensate the lumped disturbances including the errors generated by feedforward control. Finally, the stability of the proposed composite control method is proved through theoretical analysis and the effectiveness of the proposed method is validated by digital simulations.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Neural Network Feedforward Aided Composite Anti-disturbance Control for Hypersonic Morphing Vehicle

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Xingyu Wu,
  • Honglun Wang,
  • Yuebin Lun,
  • Bin Ren

摘要

In addressing the attitude tracking control problem of the hypersonic morphing vehicle (HMV), a composite control method including feedforward and feedback control is proposed in this paper. Firstly, a feedback controller is established using the backstepping method and active disturbance rejection control (ADRC) scheme to overcome the external disturbances and model uncertainties. Next, a feedforward controller based on long short-term memory (LSTM) neural network is introduced to enhance the HMV’s response speed, disturbance rejection capability and adaptability to the aerodynamic characteristics variations and uncertainties. Then, the feedback controller is redesigned to compensate the lumped disturbances including the errors generated by feedforward control. Finally, the stability of the proposed composite control method is proved through theoretical analysis and the effectiveness of the proposed method is validated by digital simulations.