WIND FARM POWER FORECAST IN MID-EASTERN COASTLINE OF CHINA

Applying 4-levels grid nested WRF (Weather Research and Forecasting) model to predict meteorological elements of wind farm field, and employing WRF model outputs and SCADA data of wind turbines to construct wind power forecast models by using the APLSR (Adapting Partial Least Square Regression) method, this paper has conducted wind power forecast in 24 hours ahead, at Wenling wind farm which located in mid-eastern coastline of China, where is subject to frequent typhoons. The results show that the WRF model has a good capability to predict wind speed variation at 1km resolution, and wind power forecasted by the constructed models agrees well with recorded data of wind farm. The MAE (Mean Absolute Error) of wind power forecast is 8.2 % over 24 hours forecast horizon.