Abstract: The cable-strut structures have the characteristics of weak stiffness and low damping, which exhibit high vibration levels under dynamic excitations. By inputting appropriate control force into the structure through the embedded actuators, an active vibration control can be achieved. The computational efficiency of active control algorithms faces more severe challenges due to the existence of objective factors such as the complexity of structural model, the diversity of controller design parameters, and the need for algorithmic robustness design. Thusly, the proper orthogonal decomposition (POD) method is proposed to achieve a data-driven reduced-order model (ROM) for improving the computational efficiency of the active control algorithm. The reduced-order space is constructed by using the structural displacement, velocity, or acceleration response data. The order of the ROM is selected automatically upon an energy-based accuracy index. The design parameters of the linear quadratic regulator (LQR) using POD-based ROM (POD-LQR) are determined according to the projection relationship, thus the control performance index does not change with the order of the ROM. The application of ROM greatly reduces the design time of POD-LQR controller. The verification example uses the Levy cable dome and 30-strut spherical tensegrity. The findings indicate that the POD-LQR controller exhibits a superior design efficiency and a vibration control effect.