电器与能效管理技术 ›› 2020, Vol. 0 ›› Issue (11): 22-28.doi: 10.16628/j.cnki.2095-8188.2020.11.004

• 研究与分析 • 上一篇    下一篇

基于EEMD-SVR模型的风电功率预测

李俊杰1, 石强1, 胡群勇2, 何立新1,3   

  1. 1.三峡大学 电气与新能源学院, 湖北 宜昌 443002
    2.广东电网有限责任公司 中山供电局, 广东 中山 528400
    3.葛洲坝电厂, 湖北 宜昌 443002
  • 收稿日期:2020-09-14 出版日期:2020-11-30 发布日期:2020-12-14
  • 作者简介:李俊杰(1996—),男,硕士研究生,研究方向为电力系统运行与优化。|石 强(1995—),男,硕士研究生,研究方向为电力系统运行与优化研究。|胡群勇(1975—),男,高级工程师,主要从事配电网络运行和管理工作。
  • 基金资助:
    国家级大学生创新创业训练计划(201711075009)

Wind Power Prediction Based on EEMD-SVR Model

LI Junjie1, SHI Qiang1, HU Qunyong2, HE Lixin1,3   

  1. 1.College of Electrical and New Energy,China Three Gorges University, Yichang 443002, China
    2.Zhongshan Power Supply Bureau of Guangdong Power Grid Inc, Zhongshan 528400, China
    3.Gezhouba Hydropower Plant, Yichang 443002, China
  • Received:2020-09-14 Online:2020-11-30 Published:2020-12-14

摘要:

风速的随机性、非线性等问题导致风电功率的预测难度较大,针对风功率的预测问题,提出一种集合经验模态分解(EEMD)和支持向量回归(SVR)模型的预测方法。首先对原始风速信号进行模态分解,由EEMD将风速信号分解为多个特征模态分量和一个残余分量,有效优化信号非线性特征;其次,依据分解得到的各分量信号训练SVR模型进而实现分量预测;最后,合并预测所得的各分量,以确定风速预测序列,并由风速与功率转换关系求得预测功率。通过案例仿真对EEMD-SVR模型的预测效果进行验证和模型对比分析,结果表明,所提模型能够实现非平稳序列的可靠分解,风电功率预测效果得到有效改善。

关键词: 风功率预测, 风速预测, EEMD, SVR, 组合模型

Abstract:

The problems of randomness and non-linearity of wind speed make it difficult to predict wind power.For the prediction of wind power,the prediction methods of ensemble empirical mode decomposition (EEMD) and support vector regression (SVR) were proposed.First,the original wind speed signal is modal decomposed,and the EEMD decomposes the wind speed signal into multiple characteristic modal components and a residual component to effectively optimize the nonlinear characteristics of the signal.Second,the SVR model is trained according to the component signals obtained by the decomposition to realize the component prediction.Finally,the predicted components are combined to determine the wind speed prediction sequence,and the predicted power is obtained from the conversion relationship between wind speed and power.Through the case simulation,the prediction effect of the EEMD-SVR model is verified and the model compared and analyzed.The results show that the model can achieve reliable decomposition of non-stationary sequences,and the wind power prediction effect is effectively improved.

Key words: wind power prediction, wind speed prediction, EEMD, SVR, combined model

中图分类号: