| 引用本文: | 史宇伟,潘学萍.计及历史气象数据的短期风速预测[J].电力自动化设备,2014,34(10): |
| SHI Yuwei,PAN Xueping.Short-term wind speed forecasting considering historical meteorological data[J].Electric Power Automation Equipment,2014,34(10): |
|
| 摘要: |
| 提出一种计及历史气象数据的短期风速预测方法。首先将历史风速数据和历史气象数据作为风速预测的原始输入，采用混合特征选择（HFS）方法对输入向量进行删选，选取与预测风速强相关的变量，生成预测模型的输入特征集；然后运用异方差高斯过程回归（HGP）模型进行建模，该模型能体现风速的随机性。根据某实测风速数据进行提前1 h风速预测，结果表明所提方法能提高风速预测精度。 |
| 关键词: 风电 风速 预测 气象数据 混合特征选择 异方差高斯过程 模型 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| Short-term wind speed forecasting considering historical meteorological data |
|
SHI Yuwei, PAN Xueping
|
|
School of Energy and Electrical Engineering，Hohai University，Nanjing 210098，China
|
| Abstract: |
| A method of short-term wind speed forecasting considering the historical meteorological data is proposed. With the historical wind speed data and historical meteorological data as its original inputs，the HFS（Hybrid Feature Selection） method is applied to select the variables strongly relevant to the wind speed and to generate the input characteristic set of the forecast model. The ＨＧＰ（Heteroscedastic Gaussian Process） model is then applied to build the forecast model for expressing the randomness of wind speed. A one-hour wind speed forecasting is carried out based on practical wind speed data and results show that the proposed method improves the accuracy of wind speed forecasting. |
| Key words: wind power wind speed forecasting meteorological data hybrid feature selection heteroscedastic Gaussian process models |
