Estimation of Wind Power Output Curve using Artificial Neural Network

Abstract

Accurate estimation of wind turbine power curve has an important role in monitoring of conditioning and controlling of wind turbines in
wind power plants. Artificial neural network (ANN) was used in this study in the prediction of horizontal axis wind turbine output power (P)
in terms of climatic data and turbine rotational speed (Ω). Artificial neural network (ANN) involved the input parameters including the wind
speed (V), atmospheric air temperature (T) and the rotational speed (Ω) of wind turbines which they were obtained from an operating power
plant. According to the derived results for the testing process, minimum mean absolute percentage of error (MAPE) and maximum correlation
coefficient (R) values were determined for an optimum rotational speed (Ω). MAPE and R values were respectively determined as 1.47%
and 0.9991 in the case of the ANN study. These results indicated well that ANN approach provided a simple and accurate forecasting in the
determination of wind turbine output power (P). Wind turbine power curve of a considered site can be rapidly predicted in a successful way
with a little error under the utilization of the ANN method when the parameters of the climatic data including the wind speed (V) and the
atmospheric air temperature (T); and as well rotational speed (Ω) of wind turbines in a wind farm are available. Thus, this method is rather
convenient during the decision stage of new wind power plant installations.