Fault-Tolerant Integrated Guidance and Control Design for Hypersonic Vehicle Based on PPO

Abstract

Aiming at the problem of the terminal guidance phase of hypersonic vehicles (HSV) under fault condition, and considering the existence of various uncertain parameters and actuator faults in the control system, a fault-tolerant integrated guidance and control design of a hypersonic vehicle based on the proximal policy optimization algorithm (PPO) is proposed. First, in view of the problem that the separate guidance and control loop design cannot make full use of the coupling relationship between the two, the relationship between the guidance loop and the control loop is considered and an integrated guidance and control system of HSV is established. Then, the integrated guidance and control problem is converted into a reinforcement learning model, the action space, state observation space, and reward function of the PPO agent are designed, and the network is initialized and designed. Simulations verify the feasibility of the proposed PPO-based IGC system.