东北大学学报:自然科学版 ›› 2016, Vol. 37 ›› Issue (2): 153-156.DOI: 10.12068/j.issn.1005-3026.2016.02.001

• 信息与控制 •    下一篇

一种参数自调整风电功率预测模型

翟军昌1, 葛延峰2, 梁鹏3, 高立群1   

  1. (1. 东北大学 信息科学与工程学院, 辽宁 沈阳110819; 2. 辽宁省电力有限公司, 辽宁 沈阳110006;3. 国网辽宁省电力有限公司 锦州供电公司, 辽宁 锦州121000)
  • 收稿日期:2014-12-12 修回日期:2014-12-12 出版日期:2016-02-15 发布日期:2016-02-18
  • 通讯作者: 翟军昌
  • 作者简介:翟军昌(1978-), 男, 辽宁东港人,东北大学博士研究生; 高立群(1949-),男,辽宁沈阳人,东北大学教授,博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61273155, 61104106).

A Parameter Self-tuning Model for Wind Power Prediction

ZHAI Jun-chang1, GE Yan-feng2, LIANG Peng3, GAO Li-qun1   

  1. 1. School of Information Science & Engineering, Northeastern University, Shenyang 110819, China; 2. Liaoning Electric Power Company Limited,Shenyang 110006,China; 3. Jinzhou City Power Supply Company, State Grid Liaoning Electric Power Company Limited, Jinzhou 121000, China.
  • Received:2014-12-12 Revised:2014-12-12 Online:2016-02-15 Published:2016-02-18
  • Contact: ZHAI Jun-chang
  • About author:-
  • Supported by:
    -

摘要: 为了提高风电功率预测精度,提出一种参数自调整风电功率预测模型.通过加权递推最小二乘(SWWRLS)方法建立预测模型,侧重当前数据对预测结果的影响,排除了历史数据对预测结果的干扰.模型通过加权递推的方法节省了存储空间,并且提高了模型对外界环境数据变化的自适应性.最后,分别采用支持向量机(SVM)方法、卡尔曼滤波(KF)方法和本文SWWRLS方法,以辽宁省某风电场的真实历史数据进行风电功率预测对比实验,实验结果表明,本文方法建立的模型具有较高的预测精度.

关键词: 风电功率预测, 最小二乘法, 滑动窗口, 支持向量机, 卡尔曼滤波

Abstract: To improve the prediction accuracy of wind power, a parameter self-tuning model is proposed. The sliding window weighted recursive least square (SWWRLS) method is used to establish the wind power prediction model, which emphasizes most recent data in the prediction, ruling out the interference of historical data. Less memory space is used in the weighted recursive model, and the adaptability of the prediction model to the external environment data is increased. Finally, the simulation is carried out on the real historical data of the wind farm in Liaoning with support vector machine, Kalman filter and SWWRLS methods, separately, and the simulation results show the superiority of the proposed method.

Key words: wind power prediction, least square method, sliding window, support vector machine, Kalman filter

中图分类号: