基于IVMD-SVR模型的短期风速预测  

Short-term Wind Speed Prediction Based on IVMD-SVR Model

作　　者：尹隆腾 陈娟[1] YIN Longteng;CHEN Juan(Wuhan University of Science and Technology,Wuhan 430070)

机构地区：[1]武汉科技大学,武汉430070

出　　处：《计算机与数字工程》2025年第2期432-436,443,共6页Computer & Digital Engineering

摘　　要：随着传统化石燃料的锐减,以及带来的环境污染的日益严重,世界各国都在重点发展作为可再生能源发电技术之一的风力发电,准确的风速预测是确保风电系统安全稳定运行的一个长期挑战。为此,提出了一种基于混沌特性来分析的变分模态分解和支持向量回归机的风速预测模型。论文提出的IVMD(Improved Variational Mode Decomposition)-SVR(Support Vector Regression)采用变分模态分解(VMD)对风速序列进行分解,降低了风速序列的复杂性和非平稳性。再通过混沌测试的方法判断分解后的时间序列是否具有混沌特性,如果具有混沌特性利用G-P算法确定嵌入维数m,如果不具有混沌特性,利用偏自相关系数确定阶数,汇总进行支持向量回归机的预测。基于真实数据的仿真验证了IVMD-SVR方法在预测性能上的显著优势,相较于其他常用方法更加高效可靠。With the sharp decline of traditional fossil fuels,as well as the increasingly serious environmental pollution,all countries in the world are focusing on the development of wind power generation as one of the renewable energy generation technolo⁃gy,accurate wind speed prediction is to ensure the safe and stable operation of wind power system is a long-term challenge.There⁃fore,a wind speed prediction model based on chaotic characteristics of variational modal decomposition and support vector regres⁃sion is proposed.The IVMD(Improved Variational Mode Decomposition)-SVR(Support Vector Regression)proposed in this pa⁃per uses variational mode decomposition(VMD)to decompose the wind speed series.The complexity and nonstationarity of wind speed sequence are reduced.Then the method of chaos test is used to judge whether the decomposed time series has chaotic charac⁃teristics.If it has chaotic characteristics,G-P algorithm is used to determine the embedding dimension m,if it does not have chaot⁃ic characteristics,partial autocorrelation coefficient is used to determine the order,and the prediction of support vector regression machine is summarized.Finally,through real data simulation experiment,the experimental results show that the IVMD-SVR meth⁃od proposed in this paper is superior to other common prediction methods in terms of prediction error,prediction accuracy and pre⁃diction effect,which proves that the prediction method is effective.

关 键 词：VMD 混沌特性 SVR 风速预测 

分 类 号：O141.4[理学—数学]