Composite Learning and Control for Nonminimum Phase Air-Breathing Hypersonic Vehicles using Online Output Redefinition

This paper investigates composite learning and control for air-breathing hypersonic vehicles, emphasizing nonminimum phase properties, parametric uncertainties, and external disturbances. First, an actor-critic prescribed deterministic learning framework is developed to acquire and store experiential knowledge with guaranteed performance, which is subsequently used for constructing an online learning controller. To address nonminimum phase behavior, an online output redefinition approach is applied to the altitude subsystem exhibiting unstable zero dynamics. Furthermore, stack-based neural disturbance observers are designed to estimate lumped disturbances, including learning approximation errors and external disturbances, whose neural weight update laws are enhanced using prediction errors. The proposed controller integrates offline learning with online control strategies, providing a significant advantage by reducing the computational burden during real-time operations. Simulation results validate the effectiveness of the proposed composite learning and control strategy.