Improved wind power prediction in connection with stochastic optimization of electricity systems

This project is one of several projects with the common purpose of predicting electricity generation from a wind farm for up to 48-60 hours based on meteorological forecasts. The project showed that 50% of the errors in wind-power predictions for the coming hour can be explained by spatial-temporal correlation where forecasts for an entire region are adjusted according to deviations in the first wind-turbine areas in the region. Moreover, the project concluded that use of several meteorological forecasts may improve wind-power predictions by 5-15%.