Wind farm power short-term prediction using WRF model and Kalman filtering

Abstract: The growing wind power industry requires more reliable power forecasting due to higher penetration of this technology in electricity generation. Due to the stochastic behaviour of wind, additional tools for improving the electricity forecast are required. This research parameterizes WRF model for wind power forecasting in Qollpana wind farm -Bolivia. Qollpana is a wind farm over a very complex terrain and 2800 meters above sea level (masl) located on the Andes. Wind power forecasting is performed for different periods of the year, such as, January, April, June, and October. Finally, the results are improved using a Kalman filter and compared with the power generated in Qollpana wind farm. The simulated WRF results of energy generated in the wind farm showed between 50 and 30 % of error. However using a Kalman filter the errors diminished up to 10 % of the energy generated by the wind farm. The time of the highest errors (at sunlight time) suggests that topographic conditions play an important role in the wind farm electricity production