Composite Learning Control for Hypersonic Flight Vehicle Using Historical Stack

Abstract

This paper proposes a composite learning control strategy to solve the control problem of hypersonic flight vehicle with model uncertainty. The strategy includes two stages, offline learning and online control. First, an adaptive neural controller is designed in the offline learning stage to learn and obtain the weight knowledge of the unknown system dynamics. Then, using the acquired empirical knowledge and the obtained histor-ical stack, an online composite learning controller is designed. Lyapunov stability analysis proves the stability of the designed controller. The proposed control strategy is based on the ideas of deterministic learning and adaptive dynamic programming, and reduces the online frequent update of neural network weights, which makes it easy to implement with good tracking performance. Finally, a comparative simulation is given to verify the effectiveness of the proposed control scheme.