Wind Speed Prediction Based on Gradient Boosting Decision Tree

As a reliable and emission-free renewable energy source, wind power plays a key role in solving the energy crisis. Accurate prediction of wind speed is one of the most significant help to rationalize the use of wind energy to alleviate the energy crisis. However, wind speed is affected by many factors, and the wind is uncertain. Therefore, in this work, we first explored the factors that affect wind speed based on Spearman. Then, in order to learn the various properties that affect wind speed, we designed the ensemble learning algorithm Gradient Boosted Regression (GDBT) to predict wind speed. The data used in the study is a climatic data set based on records from Hungary. To verify the effectiveness of our scheme, we compared our algorithm with adaptive boosting regression, support vector regression, linear regression, linear regression, and decision regression models. The experimental results show that our scheme achieves the optimal fitting performance and outstanding prediction performance.