<p>Shaking tables are widely used to simulate vibration environments for evaluating the performance, reliability, and structural integrity of various products. Among them, extremely large inertia multi-axis shaking tables are specifically designed for vibration testing of large-scale structures such as bridges and buildings. However, their complex dynamics and strong multi-stage actuator coupling present significant challenges for precise force control and motion synchronization. In this study, a nonlinear dynamic model is developed to capture the multi-stage coupling effects among actuators, inertia wheels, and the motion platform. To tackle strong coupling and synchronization challenges, a novel hierarchical decoupling strategy is introduced, integrating multi-objective optimal control and an error-sensitive state observer. This approach improves actuators synchronization and actively compensates for coupling disturbances. Experimental results demonstrate that the proposed method achieves superior decoupling performance at excitation frequencies below 0.5&#xa0;Hz, particularly in vibration testing of bridges and buildings. Compared to the iterative learning method, the proposed strategy reduces the standard deviation of actuator coupling forces by more than 20%, confirming its enhanced decoupling effectiveness under low-frequency excitation conditions. This work provides a theoretical and practical foundation for high-precision force control of large inertia multi-axis shaking tables in engineering applications.</p>