EVALUATION OF WIND PARKS OUTPUT POWER FORECAST ERROR AND WAYS TO DECREASE IT.

From a socio-economic perspective, better forecasting will reduce the total generation costs due to the more optimal dispatch of power plants. The operators of the wind parks integrated into the transmission network are responsible for presenting a 24h-forecast of their output power to the transmission system operator (TSO). The real wind power differs from the forecast one. This difference needs balancing by the rest of the energy system. In Estonian conditions, it means regulating the capacity of oil-shale-fuelled power plants, which induces an accelerated wear, additional emissions and fuel consumption of the power plants. Wind park output power is particularly difficult to forecast at wind speeds of 6 - 10 m·s-1 due to the fact that electricity generation of wind turbines changes markedly between these speeds. The most relevant metrics to measure forecast errors are Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). The absolute errors of the forecast are dependent on the forecasted wind power generation. Thus, most of the prediction problems lie in the higher end of wind output values. [ABSTRACT FROM AUTHOR]