导航

Acta Aeronautica et Astronautica Sinica ›› 2025, Vol. 46 ›› Issue (24): 632345.doi: 10.7527/S1000-6893.2025.32345

• special column • Previous Articles    

Integrated guidance and control method based on deep reinforcement learning parameter tuning

Qichao XIE, Chengyu CAO, Yiyun ZHAO, Fanbiao LI()   

  1. School of Automation Engineering,Central South University,Changsha 410083,China
  • Received:2025-06-03 Revised:2025-06-04 Accepted:2025-06-05 Online:2025-07-01 Published:2025-06-20
  • Contact: Fanbiao LI E-mail:fanbiaoli@csu.edu.cn
  • Supported by:
    Key Program of the Joint Fund for Basic Research on Large Aircraft of the National Natural Science Foundation of China(U2570207);National Science Fund for Excellent Young Scholars(62222317);Hunan Provincial Key Technology Innovation Program(2021GK1030);Key Research and Development Program of Hunan Province(2023GK2023)

Abstract:

To address the dynamic optimization problem of guidance and control parameters of hypersonic flight vehicles, a Deep reinforcement learning parameter tuning method based on the Twin Delayed Deep Deterministic policy gradient (TD3) algorithm is proposed. Firstly, the motion model of the hypersonic flight vehicle and the integrated model of guidance and control were established, and the controller based on the backstepping method was designed. The consistent final boundedness was proved via Lyapunov stability. Then, the controller parameter optimization problem was transformed into a Markov decision process model, and the data-driven online adaptive optimization of controller parameters was achieved based on the TD3 algorithm. This method constructs a parameter optimization mechanism that integrates the prior knowledge of the model and data-driven approaches, significantly enhancing the autonomous adaptability of the controller in the parameter space. Finally, the effectiveness and robustness of the proposed method were verified through numerical simulation.

Key words: hypersonic flight vehicle, integrated guidance and control, deep reinforcement learning, adaptive parameters, backstepping control

CLC Number: