上海交通大学学报(自然版) ›› 2012, Vol. 46 ›› Issue (12): 1971-1976.

• 电工技术 • 上一篇    下一篇

基于统计分析和多支持向量机的风电功率坡度事件分类预测

李福东1, 2, 吴敏1, 冯高熠3   

  1. (1. 中南大学 信息科学与工程学院, 先进控制与智能自动化湖南省工程实验室, 长沙 410083;2. 湖南省电力公司培训中心,长沙 410131; 3. 大唐华银城步新能源公司, 湖南 邵阳 422000)
  • 收稿日期:2012-06-20 出版日期:2012-12-29 发布日期:2012-12-29
  • 基金资助:

    国家自然科学基金(60974045), 国家杰出青年科学基金(60425310)资助项目

Wind Power Slope Events Classification and Forecasting Based on Statistical Analysis and Multiple Support Vector Machines

 LI  Fu-Dong-1, 2 , WU  Min-1, FENG  Gao-Yi-3   

  1. (1. School of Information Science and Engingeering, Hunan Engineering Laboratory for Advanced Control and Intelligent Automation, Central South University, Changsha 410083, China; 2. The Training Center of Hunan Electric Power Corporation, Changsha 410131, China;3. The New Energy Company of Datang Huayin, Shaoyang 422000, Hunan, China)
  • Received:2012-06-20 Online:2012-12-29 Published:2012-12-29

摘要: 为准确评估风电功率变化行为的影响,优化风电系统控制,提出了基于统计分析和多支持向量机的风电功率坡度事件分类预测方法.通过对风电功率坡度事件进行定义和分类,利用风电场的实际运行数据,对不同统计周期和不同方向的坡度事件幅度分布和时间段分布进行了统计分析,找到了功率坡度事件变化的内在规律.在此基础上,将二元支持向量机(Support Vector Machine,SVM) 拓展到多支持向量机 (Multiple Support Vector Machines,MSVMs),建立了对功率坡度事件类别的一步和多步预测.实验结果表明,所提方法具有较高的坡度事件预测精度和稳定性,可以对风电功率变化进行准确的风险预测,有利于风电系统的优化控制.    

关键词: 风电功率, 坡度事件, 多支持向量机, 类别, 预测

Abstract: To evaluate the influence of wind power fluctuations and optimize the control of wind power system, a method of wind power slope events classification and forecasting based on statistical analysis and multiple support vector machine was presented. Firstly, the wind power slope events were defined and classified. Then, the wind power data collected from a wind farm were used to investigate the classification and range of slope events, and the internal laws of slope events were explored. In this context, the binary support vector machine(SVM) was extended to multiple support vector machines(MSVMs) and was applied to the classification of slope down/up events for both one-step and multi-step ahead scenarios. Finally, the numerical results based on the wind power data verify the effectiveness of the proposed approach.  

Key words: wind power, slope event, multiple support vector machin(MSVM), classification, forecasting

中图分类号: