With the consideration of the impact of forecast errors caused by wind power uncertainty on system economic dispatch

a two-stage economic dispatch mode is proposed. On the first stage

a method based on ARIMA with a time scale of 1 hour is used to preliminarily predict the wind speed

and the predicted output values of wind power are obtained. In addition

the basic output from thermal power plants is determined with the objective of minimizing their generation costs. On the second stage

a method based on ARIMA with a time scale of 15 minutes is used to preliminarily predict the wind speed. a intra-day realtime optimal dispatching model with the time scale of 1-hour cycle and with the objective of minimizing the frequency deviation and net-work loss is established to determine the actual output. Thinking about the uncertainty of wind power

the second stage is based on the unit output of the first stage. In addition

at this stage

the dynamic power flow considering wind power frequency modulation is introduced into economic dispatch in order that frequency quality can be considered. The accelerating power in the system is shared by wind

fire

and load in a certain proportion. According to the characteristics of frequencymodulation of wind power related to wind speed and load shedding level

different equivalent static adjustment coefficients can be set through overspeed or pitch adjustment under different wind speeds to better play the role of wind power participating in frequency modulation to share unbalanced power. Finally

a multi-objective solution is established based onthe Pareto curve formed by the particle swarm optimization algorithm

and an example of improved IEEE14-node system is simulated to verify the proposed method.