Wind Power Forecasting Based on Time Series and Neural Network

The wind farm output power have the characteristics of dynamic, random, large capacity etc, which brought great difficulty for incorporating the wind farm in the bulk power system. In order to rationally regulate the power supply system in large grid connected wind power system and reduce the spinning reserve capacity of the power supply system and operating costs, it is necessary to forecasting the capacity of wind power. For the randomness of the wind farm output, we use the ARMA (q, p) model of time series to forecast wind speed and atmospheric pressure, and using the RBF neural network based on this to forecast wind power. Taking the data of measured wind speed and atmospheric pressure from a wind farm as example, to validate the method described above, and the result show that the method has a certain practicality.