Photovoltaic and other renewable energy sources, along with energy storage systems, are integrated into microgrids through power electronic interfaces. This results in a significant low-inertia issue, posing severe frequency stability risks. To enhance the frequency dynamic performance of microgrids, VSG(virtual synchronous generator) control strategy based on CMPC(cascaded model predictive control) was proposed, leveraging MPC(model predictive control) to collaboratively optimize both the outer and inner loops. In the outer loop, the rolling optimization capability of MPC was utilized to dynamically adjust the active power reference input to the VSG in real time. In the inner loop, DSVM-MPC(discrete space vector model predictive control) method was employed, which optimizes the selection of the optimal VV(voltage vector) through a two-stage process. This approach minimizes tracking errors from the outer loop output, leading to more precise frequency control. Simulation results demonstrate that the proposed strategy improves frequency response and current limiting performance while effectively preventing frequency overshoot.