To solve the problem that wind farms equipped with energy storage systems are difficult to balance the grid peak shaving demand and their own operational efficiency under the condition of limited local information，a two-stage optimal scheduling model for wind-storage combined systems with localized information is proposed.In the day-ahead optimal scheduling stage，based on the easily obtainable total system load curve，the charging and discharging power of the energy storage is actively optimized with the goal of minimizing the peak-valley difference of the net load at the grid connection point；under the premise of not relying on global information，the anti-peak shaving characteristics of wind power are improved to indirectly assist the grid in peak shaving.In the intraday rolling optimal scheduling stage，based on ultra-short-term prediction data，the power adjustment is rollingly optimized to smooth the wind power fluctuation with the goals of minimizing the day-ahead plan deviation and the wind curtailment，so as to realize the time-sharing reuse of energy storage on two time scales of macro peak shaving and micro suppression.In addition，an improved beetle swarm algorithm （IBSO） integrating chaotic opposition-based learning and Lévy flight is proposed to solve the difficulty caused by the high-dimensional multi-period coupling constraints of the model.