Because of the intermittent and uncertainties of wind speed, the difficulty of absorbing wind power increases with wind power penetration growing continuously. The prediction error of wind farm output is large, and it varies with the wind speed and fluctuation, therefore, different reserve capacity is needed to meet power supply and demand balance. Due to above issue, this paper presents a gradual optimization of day-ahead and within-day dispatching schedule. The wind power and predication error band is involved during the implementation of dispatching schedule, and the dispatching schedule of remaining time is adjusted according to the newest predication message to improve the accuracy of dispatching schedule gradually, and the adjustment includes the output power and reverse capacity of conventional generators. Compared to the implementation of dispatching schedule base on general prediction information, this model can reflect the impact of wind power predication error on dispatching schedule more effectively, and the reserve capacity can be adjusted accurately. The conclusion is demonstrated with the simulation on IEEE 30-bus system.