基于改进经验模态分解和支持向量机的风电功率预测研究

王涛; 高靖; 王优胤; 史喆; 刘涛; 杨博; 严风

王涛,高靖,王优胤,史喆,刘涛,杨博,严风.基于改进经验模态分解和支持向量机的风电功率预测研究[J].电测与仪表,2021,58(6):49-54.
Wang Tao,Gao Jing,Wang Youyin,Shi Zhe,Liu Tao,Yang Bo,Yan Feng.Wind%20Power%20Prediction%20Based%20on%20Improved%20Empirical%20Mode%20Decomposition%20and%20Support%20Vector%20Machine[J].Electrical Measurement & Instrumentation,2021,58(6):49-54.

基于改进经验模态分解和支持向量机的风电功率预测研究

Wind%20Power%20Prediction%20Based%20on%20Improved%20Empirical%20Mode%20Decomposition%20and%20Support%20Vector%20Machine

DOI：10.19753/j.issn1001-1390.2021.06.007

中文关键词: EMD 采样率问题支持向量机风电预测

英文关键词:EMD %20Sampling%20Rate%20Problem %20Support%20Vector%20Machine %20Wind%20Power%20Prediction

基金项目:国家电网公司科技项目

作者中文名	作者英文名	单位
王涛	Wang Tao	国网辽宁省电力有限公司经济技术研究院
高靖	Gao Jing	国网辽宁省电力有限公司经济技术研究院
王优胤	Wang Youyin	国网辽宁省电力有限公司经济技术研究院
史喆	Shi Zhe	国网辽宁省电力有限公司经济技术研究院
刘涛	Liu Tao	国网辽宁省电力有限公司经济技术研究院
杨博	Yang Bo	国网辽宁省电力有限公司经济技术研究院
严风	Yan Feng	华北电力大学电气与电子工程学院

摘要点击次数: 1539

中文摘要:

风电大规模接入给电网安全运行带来了较大的挑战,风电预测是主要的解决措施之一。针对风电功率信号的非线性和非平稳性特征,提出了一种基于改进经验模态分解(IEMD)和支持向量机(SVM)的风电信号组合预测方法。首先,阐述了EMD的基本原理和优缺点,针对其分解非线性非平稳信号时的采样率问题,提出了一种消减欠冲现象的改进方法。其次,以辽宁某风电场提供的风电数据为研究实例,利用IEMD将风电信号分解为一组较为平稳的子序列分量；然后,运用SVM理论分别构建风电信号低频和中频分量的EMD-SVM和IEMD-SVM组合预测模型,并在MATLAB中仿真对比了两种模型的预测结果。研究结果表明,IEMD-SVM的组合预测模型在分解风电信号时能够有效减少欠冲点数目,较好地表现原信号的整体趋势,与EMD-SVM预测方法相比具有更高的精度和准确度。

英文摘要:

Large-scale%20wind%20power%20integration%20brings%20great%20challenges to the%20safe%20operation%20of%20power%20grid,%20and%20wind%20power%20forecasting%20is%20one%20of%20the%20main%20solutions.%20Considering%20the%20non-linearity%20and%20non-stationarity%20of%20wind%20power%20signals,%20a%20combined%20forecasting%20method%20of%20wind%20power%20signals%20based%20on%20improved%20empirical%20mode%20decomposition%20(IEMD)%20and%20support%20vector%20machine%20(SVM)%20is%20proposed.%20Firstly,%20the%20basic%20principle%20and%20advantages%20and%20disadvantages%20of%20EMD%20are%20described.%20Aiming%20at%20the%20problem%20of%20sampling%20rate%20when%20EMD%20decomposes%20non-linear%20and%20non-stationary%20signals,%20an%20improved%20method to reduce%20under%20impulse%20is%20proposed.%20Secondly,%20taking%20wind%20power%20data%20provided%20by%20a%20wind%20farm%20in%20Liaoning%20Province%20as%20an%20example,%20wind%20power%20signals%20are%20decomposed%20into%20a%20set%20of%20relatively%20stable%20subsequence%20components%20using%20IEMD.%20Then,%20EMD-SVM%20and%20IEMD-SVM%20combined%20forecasting%20models%20of%20low-frequency%20and%20intermediate-frequency%20components%20of%20wind%20power%20signals%20are%20constructed%20respectively%20by%20using%20SVM%20theory,%20and%20the%20forecasting%20results%20of%20the%20two%20models%20are%20simulated%20and%20compared%20in%20MATLAB.%20The%20results%20show%20that%20IEMD-SVM%20combined%20forecasting%20model%20can%20effectively%20reduce%20the%20number%20of%20undershoots%20when%20decomposing%20wind%20power%20signals,%20and%20better%20represent%20the%20overall%20trend%20of%20the%20original%20signal.%20Compared%20with%20EMD-SVM%20forecasting%20method,%20it%20has%20higher%20accuracy%20and%20accuracy.

查看全文查看/发表评论下载PDF阅读器

关闭