Integration of offsite wind speed measurements in shortest-term wind power prediction systems

Shortest-term wind power forecasts of an individual wind farm are normally based on numerical weather predictions (NWP) and respective power measurements of the recent past. In this study offsite wind measurements has been investigated to improve the shortest term forecast of wind power and to setup an alternative forecast system if measured wind power from the predicted wind farm is temporal not available. Especially at high wind speeds the offsite wind measurement provide additional information for the forecast system, due to the fact that the constant power output of the wind farm running at nominal power provide no information for the forecast system. Wind speed measurements from the ISET wind measurement network are investigated with respect to additional information content concerning the forecast of a single wind farm. Based on a correlation analysis between offsite wind and wind farm power measurements depending on spatial distance and time lag an individual offsite wind measurement has been allocated for each wind farm. The corresponding wind measurement is included in the neural network based forecast system from the ISET.As expected the study reveals a minor improvement of the global forecast error but a significant one at higher wind speed levels. However the exclusive use of offsite wind measurements only in combination with NWP-models leads to a satisfying forecast quality and shows the ability to substitute forecasts based on recent power measurements in case of breakdown.