关键词: 轨道博弈, 执行器故障, 容错控制, 自适应估计, 意图识别, ε-纳什均衡

Abstract: To address the issues of unknown attack intent and actuator faults in the defender spacecraft within a target-attacker-defender three-party game, this paper proposes a cooperative fault-tolerant control method that integrates adaptive estimation and differential game theory to achieve efficient protection of the target. Two typical scenarios are considered: the target moving along an arbitrary trajectory and the target actively evading. First, an adaptive estimation law is designed using adaptive estimation theory to online identify actuator faults in the defender spacecraft. On this basis, composite learning laws are constructed for the two scenarios to enable the defender spacecraft to learn and predict the control gain matrix of the attacker online, and subsequently tailor the optimal fault-tolerant control strategy for the defender. The proposed strategy only requires interval excitation condition rather than persistent excitation, thereby reducing computational burden and saving onboard resources. Theoretical analysis shows that the designed three-party game strategy achieves an -Nash equilibrium, ensuring approximate equilibrium under the influence of estimation errors. Finally, simulation results demonstrate that the proposed fault-tolerant game method can effectively predict the attacker's intent and perform interception, thereby protecting the target spacecraft.

Key words: Orbital game, actuator fault, fault-tolerant control, adaptive estimation, intent recognition, ε-Nash equilibrium