A combination prediction model for wind farm output power

Wind power´s volatility and intermittence have a profound impact on power system´s security and economic operation. However, high-precision power prediction is the important prerequisite to reduce the influence of wind power on the power system. This paper illustrates a wind power prediction model based on time-series and back propagation artificial neural network (BP-ANN), considering wind speed, temperature, humidity, geographical conditions and other factors. Taking account of approximate linear relationship between wind speeds, the prediction model of wind speed was built based on time-series, and the model of wind speed-to-power was set up in the way of the nonlinear mapping relationship based on the method of BP-ANN. The paper predicts wind power based on the measured data of 24h ahead. By analyzing predicted data, it shows that the combined prediction model based on time-series and BP-ANN is effective.