Multi-step wind speed estimation based on artificial neural network using secondary separation technique

In recent years, the importance of integrating the production of wind energy into electrical energy networks has been increasing rapidly. The biggest challenge to integrate wind energy into the power grid wind power is variability and discontinuity. To deal with this situation, the best approach is to predict future values of wind power production. Wind speed estimation methods with high accuracy are an effective tool that can be used to minimize these problems. In this study, a hybrid estimation system consisting of 4 modules as data preprocessing, clustering, estimation and evaluation has been proposed as a model. The meteorological parameters used in the study were obtained from the wind measurement station build in the campus of Engineering and Nature Science Faculty in Gaziosmanpaşa University. The wind speed was measured in the 10 minutes intervals via cencors used in this station and one-year data was included in the analysis. The proposed model is based on decomposition techniques for the elimination of high frequency signals to reduce the effect of noise in the raw data series and to extract features in data preprocessing, the hurst exponential coefficient in order to extract the similarity property in the wind speed data in the data clustering modüle, prediction models involving back propagation artificial neural networks and the latest research findings (MAPE, RMSE, R²) published in high grade and prestigious. The results of the analysis perform better than the traditional prediction models (EEMD-VDM-BPNN and EEMD-EWT-BPNN) compared in terms of predictive accuracy of the proposed innovative hybrid model. When the MAPE values obtained in the proposed hybrid model are compared, they are decreased by 41.16% and 78.80%, respectively, compared to those obtained in traditional models.