影响因子：2.2
DOI码：10.1080/15567249.2025.2468687
所属单位：哈尔滨工程大学船舶工程学院
发表刊物：Energy Sources, Part B: Economics, Planning, and Policy
刊物所在地：美国
项目来源：Key Research Project in Universities of Henan Province [No.24B480012, No.25A450004]; Key Specialized
关键字：Ensemble empirical mode decomposition (EEMD); support vector regression (SVR); wavelet neural network (WNN); smart grid; short-term electric load forecasting
摘要：Accurate power load forecasting is very important to optimize energy distribution and maintain the stability of power grid. Traditional models are often difficult to cope with complex load fluctuations, resulting in inaccurate forecasts. This paper proposes a hybrid forecasting model EEMD-SVR-WNN, which integrates empirical mode decomposition (EEMD), support vector regression (SVR) and wavelet neural network (WNN). EEMD decomposed the power load series with large fluctuations into multiple frequency components, then forecasted the high-frequency components according to the nonlinear characteristics of SVR, and forecasted the medium frequency components by using the self-learning ability of wavelet neural network. Applied to the load data of New York City, the forecast accuracy and peak value of the model are better than the traditional methods. Compared with single model and other combined models, EEMD-SVR-WNN model has better stability, interpretability and higher precision.
备注：Key Research Project in Universities of Henan Province [No.24B480012, No.25A450004]; Key Specialized Research and Development Breakthrough Program in Henan Province [No.242102240051]
合写作者：Hui-Zhen Wei,黄信博
第一作者：范国峰
论文类型：期刊论文
通讯作者：洪维强
论文编号：2468687
学科门类：工学
文献类型：J
卷号：20
ISSN号：1556-7249
是否译文：否
发表时间：2025
收录刊物：SCI
发布期刊链接：https://www.tandfonline.com/doi/full/10.1080/15567249.2025.2468687