Reinforcement learning adaptive risk-sensitive fault-tolerant IGC method for a class of STT missile

The paper focuses on the reinforcement learning adaptive risk-sensitive fault-tolerant integrated guidance and control (IGC) for a class of skid-to-turn (STT) missile with non-affine characteristics, stochastic disturbances and unknown uncertainties, actuator faults, and the risk-sensitive index. The initial focus of this study is introducing an extended integration system, which aims to address the complex control problem caused by the non-affine structure of the control signal. Considering stochastic disturbances and unknown uncertainties, a novel adaptive actor-critic design is proposed, which aims to guarantee the input-state stability of the system. Following this, hyperbolic tangent functions and adaptive boundary estimation techniques mitigate the effects of disturbance-induced jitter and actuator fault-induced bias inside the control system. The cost standard of the risk-sensitive index that we desired can be made arbitrarily minimal by solving a specific inequality. Moreover, the suggested control strategy enhances the interception capabilities of the missile when engaging maneuvering targets and mitigates the excessive caution inherent in current adaptive robust control approaches. The stability of the non-affine integrated guidance and control (NAIGC) closed-loop system is demonstrated using the Lyapunov theory. Finally, numerical simulations are conducted to verify the efficacy and superiority of this method.