Data-Driven Modeling and Control of Tethered Satellite Dynamics Based on the Koopman Operator
Zhifeng Liao, Ti Chen
Abstract: Due to the lightweight and highly flexible structure, the tethered satellite system exhibits nonlinearity which makes it difficult to design controllers. This paper applies Koopman operator theory by selecting Koopman operator observable function for the tethered satellite dynamic system and utilizing a data-driven approach to linearize the system globally. The data-driven model can account for unmodeled disturbances affecting the system, and the resulting model is linear, which facilitates controller design. In this study, numerical simulations of the tethered dumbbell model are conducted to obtain the data required for data-driven modeling. Based on the selected observable functions, the tethered satellite system is linearized using the extended dynamic mode decomposition algorithm. With the same initial conditions and control inputs, compared with the original nonlinear model, the state trajectory errors of the linearized model are acceptable. The results demonstrate the feasibility of using the data-driven approach to linearize the nonlinear model. Finally, based on the obtained linear model, an LQR controller is designed to achieve the stable deployment of the tethered satellite and keep the pitch angle and roll angle near 0° at steady state. The results show that the designed controller effectively achieves the desired performance.
文章链接：https://www.sciencedirect.com/science/article/pii/S2405896325020750
