Resilient Predefined-Time Neuroadaptive Control for Spacecraft Formation: Conditional Compensation and Hierarchical Learning

This article presents a predefined-time neuroadaptive control strategy for a six-degree-of-freedom spacecraft formation system, capable of addressing communication link failures (CLFs) and lumped disturbances. A resilient predefined-time distributed observer is initially devised to reconstruct the leader’s spatial states under CLF constraints. Subsequently, a conditional disturbance compensation scheme based on proportional quantization is formulated. It not only refines the characterization of beneficial disturbances but also attenuates detrimental effects of observation errors during attribute transitions. Then, a predefined-time neuroadaptive hierarchical controller is proposed to handle potential failures in neural learning. Its distinctive feature lies in the integration of a data-driven weight update mechanism with a regional partitioning of learning attributes, which markedly elevates learning capabilities. Global stability is rigorously guaranteed via Lyapunov theory. The effectiveness and engineering applicability of the proposed findings are corroborated through numerical simulations and full-physical ground experiments.