| 引用本文: | 李慧杰,刘亚南,卫志农,李晓露,Kwok W Cheung,孙永辉,孙国强.基于相关向量机的短期风速预测模型[J].电力自动化设备,2013,33(10): |
| LI Huijie,LIU Yanan,WEI Zhinong,LI Xiaolu,Kwok W Cheung,SUN Yonghui,SUN Guoqiang.Short-term wind speed forecasting model based on relevance vector machine[J].Electric Power Automation Equipment,2013,33(10): |
|
| 摘要: |
| 通过对风速的时间序列进行分析，表明该序列具有混沌特性。在此基础上，利用相空间重构理论建立基于相关向量机（RVM）的短期风速预测模型，并对不同的核函数进行分析，选出最优的核函数。与现有的风速预测模型相比，该模型具有高稀疏性、核函数选择灵活等优点。仿真结果表明，与BP神经网络和支持向量机（SVM）模型相比，RVM模型预测精度更高。 |
| 关键词: 神经网络 支持向量机 相关向量机 相空间重构 短期风速预测 模型 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金资助项目（51277052，51107032， 61104045） |
|
| Short-term wind speed forecasting model based on relevance vector machine |
|
LI Huijie1, LIU Yanan2, WEI Zhinong2, LI Xiaolu1, Kwok W Cheung3, SUN Yonghui2, SUN Guoqiang2
|
|
1.ALSTOM Grid Technology Center Co.，Ltd.，Shanghai 201114，China;2.Research Center for Renewable Energy Generation Engineering，Ministry of Education，
Hohai University，Nanjing 210098，China;3.ALSTOM Grid Inc.，Redmond，Washington 98052，USA
|
| Abstract: |
| Analysis of chronological wind speed series shows its chaos，according to which，a short-term wind speed forecasting model based on RVM（Relevance Vector Machine） is built by phase space reconstruction and its optimal kernel function is chosen by kernel function analysis. Compared to existing wind speed forecasting models，it has higher sparseness，as well as higher flexibility in kernel function selection. Simulative results show that，its forecast accuracy is higher than those of BP neural network and SVM（Support Vector Machine）-based model. |
| Key words: neural networks support vector machines relevance vector machine phase-space reconstruction short-term wind speed forecasting models |
