Comparison between ANNs and linear MCP algorithms in the long-term estimation of the cost per kWh produced by a wind turbine at a candidate site: A case study in the Canary Islands

Abstract

In the work presented in this paper Artificial Neural Networks (ANNs) were used to estimate the long-term wind speeds at a candidate site. The specific costs of the wind energy were subsequently determined on the basis of the knowledge of these wind speeds. The results were compared with those obtained with a linear Measure-Correlate-Predict (MCP) method. The mean hourly wind speeds and directions recorded over a 10 year period at six weather stations located on different islands in the Canary Archipelago (Spain) were used as a case study. The power-wind speed curves for five wind turbines of different rated power were also used. The mean absolute percentage error (MAPE), Pearson's correlation coefficient and the Index of Agreement (IoA) between measured and estimated data were used to evaluate the errors made with the different metrics analysed.Amongst the conclusions that can be drawn from the study it can be stated that, in all the cases analysed, the MAPE of the specific cost of the energy obtained using ANNs was lower than that obtained with the linear MCP method that was employed. In some cases reductions in the error of up to 26.5% were achieved with the ANNs with respect to the method used for purposes of comparison. However, when the correlation coefficients between the short-term data series of the candidate and reference sites were relatively low, the MAPEs generated by all the analysed metrics were higher when the estimation models were used than when the data for the short-term period at the candidate site was taken as being representative of the long-term wind performance at that site. (C) 2011 Elsevier Ltd. All rights reserved.