Hierarchical fault-tolerant control for over-actuated hypersonic reentry vehicles

This paper proposes a performance-focused and deep learning-based hierarchical fault-tolerant control framework for over-actuated hypersonic reentry vehicles. The hierarchical fault tolerance mechanism refers to the fault tolerance that can be realized both in the control allocation and control layers. First, a long short-term memory neural network-based fault diagnosis unit is proposed to solve the difficult problem of diagnosing actuator fault information under multisource disturbances. Based on the diagnosed fault information, the robust least-squares control allocation method is deployed to distribute the desired control moment to the actuator and achieve fault tolerance in the control allocation layer. Then, combined with a fixed time extended state observer, an improved multiple-degree-of-freedom prescribed performance control method is used to compensate the lumped disturbances, including control allocation error, ensure the prescribed tracking performance, and realize fault tolerance in the control layer. Finally, the theoretical analysis proves the stability of the proposed hierarchical fault-tolerant control scheme, and the effectiveness is verified by numerical simulation.