Multi-layer cooperative combined forecasting system for short-term wind speed forecasting

Abstract

Short-term wind speed forecasting is crucial to enhance the operational efficiency and increase the economic benefit of wind power generation systems. A substantial number of studies were conducted to enhance wind speed forecasting. Combined models are increasingly being preferred, as they always yield better results. However, the current combined models either do not fully consider the different characteristics hidden in the wind speed series or are restricted by the difficulties in modeling very unstable and irregular data, which leads to a limited forecasting enhancement. To overcome the limitation of the current combined models and provide a more reliable wind speed prediction, this study proposes an efficient combined forecasting system that innovatively adopts a more advanced multi-layer cooperative combined strategy, which inherits the strengths of the conventional combined strategies while simultaneously avoiding some of their limitations. It successfully overcomes some drawbacks of some current models, for example, low accuracy and high instability, and significantly improve the forecasting performance. Experiments conducted in this study demonstrate that the proposed system is superior to other single or combined methods, as it can always obtain a higher forecasting accuracy and stability. It can offer a practical method to forecast wind speed.