风资源超短期预估中的多数据源降维预处理方法研究  

Multiple Data Source Dimensionality Reduction Pretreatment Used in Ultra-Short Term Wind Resource Forecast

作　　者：鲁宗相[1] 李剑楠[1] 乔颖[1] 赵俊屹[2] 杨超颖[2] 

机构地区：[1]电力系统及发电设备控制和仿真国家重点实验室(清华大学),北京市海淀区100084 [2]国网山西省电力公司,山西省太原市030001

出　　处：《电网技术》2015年第5期1275-1280,共6页Power System Technology

基　　金：国家863高技术基金项目(2011AA05A103);国家自然科学基金项目(51190101)~~

摘　　要：准确预估自然状态下的风资源是进行风电优化调度的基础。为了充分利用多数据源实现有用信息的筛选和提取来提高预估精度,同时有效控制数据规模和复杂性,在风资源超短期预估中引入数据预处理环节是必要的。因此提出了基于复相关系数的数据筛选方法及基于典型相关分析的序列降维方法,构建了多维到1维序列映射模型用于多数据源的质量提升和降维简化,作为前置数据处理环节纳入到基于遗传算法和反向传播(back propagation,BP)神经网络的风资源超短期预估方法中。最后通过实际算例证明了该数据预处理方法在提高预估精度方面具有显著的效果。It is the foundation of power optimal dispatching to forecast the wind resource under natural state accurately. To implement the screening and extraction of useful information by making full use of multiple data resource to improve the accuracy of forecasting and in the meanwhile to control the scale and complexity of input data,, it is necessary and indispensable to lead the data preprocessing into ultra-short term forecasting of wind resource. A multiple correlation coefficient based data screening method and a canonical correlation analysis based dimensionality reduction method are proposed, and a model, which maps the multidimensional sequence to one-dimensional sequence, is constructed to applied in both quality improvement and dimensionality reduction of multi-data source and the model is regarded as a prepositive data-processing step and brought into the ultra-short term forecasting of wind resources based on genetic algorithm and BP neural network. Finally, the effect of the proposed data pretreatment method in the aspect of improving forecasting accuracy is validated by case study of actual wind farms.

关 键 词：风资源 多数据源 降维技术 数据筛选 预估方法 

分 类 号：TM614[电气工程—电力系统及自动化]