Data-Driven Geometric Control Method to the Attitude Control of a Carrier Rocket

The focus of this paper is to obtain a data-driven model of the rocket using a manifold-based approach and designing the controller for closed-loop rocket dynamics system. To meet the requirements of the optimal model, appropriate orthogonal decomposition is utilized to select low-order subspaces, and the state space model is fitted to the subspaces using the least squares method. On this basis, iterative algebraic equations for the construction of optimal feedback under Hamiltonian cohomology conditions are derived by designing gradient-based minimisation algorithms in convex positive definite matrix manifold functions. Finally, we demonstrate the reliability of the proposed manifold optimisation algorithm for a rocket attitude control model.