1. Weather division‐based wind power forecasting model with feature selection
- Author
-
Liu Kaipei, Xiong Yindi, and Qin Liang
- Subjects
Wind power ,Renewable Energy, Sustainability and the Environment ,business.industry ,Computer science ,020209 energy ,020208 electrical & electronic engineering ,Stability (learning theory) ,Wind power forecasting ,Feature selection ,Sample (statistics) ,02 engineering and technology ,Division (mathematics) ,computer.software_genre ,Hierarchical clustering ,Term (time) ,0202 electrical engineering, electronic engineering, information engineering ,Data mining ,business ,computer ,Physics::Atmospheric and Oceanic Physics - Abstract
The sample division-based hybrid model is an enforceable approach to improve wind power forecasting accuracy in the short term. These models up to now prefer to keep the input same for all the individual schemes, which weaken the effort of division and restrict the further improvement of the accuracy. To this end, a weather division-based wind power forecasting model with ensemble feature selection is proposed for refinement. The methodology comprises three stages: the division of wind power associated weather based on hierarchical clustering with the DTW distance metric, ensemble feature selection framework considering both predictive accuracy and stability, and wind power prediction based on machine learning algorithms for each weather type. As a test case, the proposed methodology is applied to the data of a wind farm group in Northwest China. With respect to the single models, the proposed method has improved the predictive accuracy by up to 30% at three error metrics, and the weather associated features are discussed.
- Published
- 2019
- Full Text
- View/download PDF