The intermittency of wind power fluctuations and the low reliability of reporting scheduling plans have always affected the friendly access of grid connected wind power. “One machine, one storage” provides an effective solution. An innovative control method for dual-frequency PWM pulses in an improved bidirectional DC/DC Zeta Sepic circuit model is proposed to address the controllability of energy throughput in energy storage systems with dynamic deviation compensation for wind power planning. Firstly, by analyzing the working timing logic mechanism of the energy transfer converter, the feasibility of the energy storage throughput circuit is verified. Then, by analyzing the small signal control model of the bidirectional DC/DC converter during charging and discharging operation, the dual-frequency PWM automatic control of charging and discharging conditions is achieved, and the controllability of power exchange of the energy storage system at the wind turbine grid connection point is achieved. Finally, the model predictive control (MPC) method is adopted to effectively compensate the deviation of wind power plans, which can be accurately tracked. The simulation experiment demonstrates the effectiveness and feasibility of the converter model structure and control algorithm.