Developing a hybrid method based on extreme learning machine and wavelet package transform for short-term wind speed forecasting

Wind energy has a variable characteristic and disrupts the supply-demand balance in electrical grids. Wind energy forecasts are needed to mitigate this disruptive effect. Due to the cubic relationship between wind energy and wind speed, wind speed forecasts play a crucial role in wind energy forecasting process. In this thesis study, a hybrid method that combines the wavelet packet transform (WPT) and extreme learning machines (ELM) is proposed for wind speed forecasting. In the proposed method, the wind speed is decomposed into components using WPT, and an ELM-based model is developed for forecasting each component. Finally, the forecasted speed components are algebraically summed to obtain the wind speed forecast. The experimental studies of this thesis were conducted using field measurements taken from the province of Osmaniye. The developed wind speed forecasting model was evaluated using performance metrics such as correlation coefficient (R), root mean square error (RMSE), and mean absolute error (MAE). The performance tests realized in this thesis study revealed that the proposed wind speed forecasting method achieved the values of 0.0218, 0.0155, and 0.9901 for RMSE, MAE, and R performance metrics, respectively, at a 1-hour forecasting horizon. Based on the findings of the thesis, it can be concluded that the proposed method can forecast wind speeds with high quality and can be successfully utilized in the process of wind energy forecasting.