A novel combined framework for short-term wind speed forecasting based on data preprocessing with sequence reconstruction and Grey Wolf optimization

The forecasting of wind speed in the short term is crucial for the production of wind power and greatly influences control and operational choices. Many prediction techniques have been developed to increase the accuracy of wind speed predictions. However, existing forecasting techniques often overlook the significance of data decomposition and are susceptible to various constraints inherent in traditional individual models, which can lead to suboptimal forecasting accuracy. This study develops a combined forecasting system using data denoising, an ensemble strategy, various classical forecasting models, and an optimized algorithm. More particularly, in order to validate the performance of the proposed combined forecasting system, the original 10-minute wind speed sequence from a wind farm in Penglai, China, is used in this study. The experiment’s results and debate show that the proposed combined forecasting system has improved forecasting accuracy as compared to classical individual forecasting models.