Short-term Wind Speed Prediction with Ensemble Algorithm

Wind energy is a crucial component of renewable energy. Accurate prediction of wind speed will contribute a lot to the energy generation. This paper uses the data of wind speed in Belnoit, USA, studying the methods of predicting short-term wind speed and its direction. Firstly, different machine learning methods were compared to obtain initial solution, then the hybrid models using different ensemble learning algorithm also proposed. To valid the prediction accuracy and stability, various types of evaluation indicators were listed and calculated. The results shows that different models could capture different information, so we need to use several evaluation indicators to have a more exhaustive image about each model. Moreover, a better initial prediction with single model will allow us have a general result in prediction at first. Finally, ensemble learning could combines advantages of each model and reduce errors in the prediction process.