Skip to content
Open access

Physics-Driven Koopman Lifting and Adaptive Optimal Control for Rigid-Flexible Spacecraft Attitude Maneuvering

2026 · IEEE Transactions on Automation Science and Engineering · Vol 23, pp. 18259-18275 · 0 citations · 55 references

Abstract

To address the inherently strong nonlinearities and rigid-flexible coupled dynamics in flexible spacecraft, this paper proposes an analytical Koopman-operator-based adaptive optimal control strategy for attitude maneuvers and active vibration suppression. A novel physics-driven Koopman global linear lifting approach is developed through a set of analytical observable functions derived from system dynamics, rather than relying on conventional data-driven approaches such as dynamic mode decomposition (DMD) or extended dynamic mode decomposition (EDMD). This analytical lifting transforms the nonlinear dynamics into an exact global infinite-dimensional linear form, from which finite-dimensional representations for both attitude stabilization and tracking are established with high approximation accuracy and superior physical interpretability. Leveraging the induced linear structure, an efficient adaptive value iteration optimal control algorithm is formulated for attitude maneuvering without solving complex nonlinear Hamilton-Jacobi-Bellman (HJB) equations or repeated online Riccati computations. This approach significantly reduces the online computational burden compared to traditional nonlinear optimal control and Riccati-based online schemes, while ensuring superior control accuracy over conventional local linearization techniques. The effectiveness, computational efficiency, and practical advantages of the proposed strategy are validated through a high-fidelity Python/MuJoCo-based 3D simulator of flexible spacecraft attitude maneuvers under complicated scenarios. Note to Practitioners—Advanced missions such as remote sensing and deep space exploration increasingly utilize flexible spacecraft equipped with large-scale flexible structures, which face critical challenges from strong nonlinearities and rigid-flexible coupled dynamics. Conventional control approaches often exhibit limitations: local linearization techniques frequently lose accuracy during large-angle maneuvers, while nonlinear optimizations rely on complex models of the flexible spacecraft that impose heavy computational burdens and fail to achieve rapid real-time adjustments on-board. To overcome these limitations, this paper proposes a fast adaptive optimal control method based on an analytically derived Koopman operator for flexible spacecraft attitude systems. The resulting physics-derived global linear representation enables efficient value-iteration-based controller learning, improves control accuracy over local linearization methods, and provides favorable computational efficiency for potential on-orbit implementation. Beyond spacecraft, this physics-driven approach is potentially applicable to a broader class of nonlinear systems with rigid-flexible coupling, such as robotic manipulators and precision machinery with elastic components.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.