基于VMD和LSTM的超短期风速预测  

Ultra-short-term wind speed prediction based on VMD-LSTM

作　　者：王俊 李霞[1] 周昔东 张凯 WANG Jun;LI Xia;ZHOU Xidong;ZHANG Kai(The College of Hohai,Chongqing Jiaotong University,Chongqing 400074,China;The College of Shipping and Ship Engineering,Chongqing Jiaotong University,Chongqing 400074,China)

机构地区：[1]重庆交通大学河海学院,重庆400074 [2]重庆交通大学航运与船舶工程学院,重庆400074

出　　处：《电力系统保护与控制》2020年第11期45-52,共8页Power System Protection and Control

基　　金：重庆市科委社会民生专项基金资助(cstc2016shmszx30002)。

摘　　要：风速具有非线性、非平稳性以及随机性等特点。为提高超短期风速预测精度,提出一种基于变分模态分解(VMD)和长短期记忆网络(LSTM)的超短期风速预测新方法。首先利用变分模态方法将风速序列分解成一系列不同的子模态以降低原始数据的复杂度和非平稳性对预测精度的影响。再对得到的风速子模态分别建立LSTM模型,进行超前1步风速预测。最后叠加各子模态的预测结果得到最终预测风速。对比分析结果显示,该模型的预测精度优于其他多种典型风速预测模型,该模型在超短期风速预测方面表现出较好的性能。Wind speed has characteristics of non-linearity,non-stationarity and randomness.In order to improve the accuracy of ultra-short-term wind speed prediction,a new method based on VMD and LSTM is proposed.First,the VMD is used to decompose the wind speed sequence into a series of IMF to reduce the complexity and non-stationarity of the original data.Secondly,LSTM models with 1-step ahead wind speed prediction are established for each IMF.Finally,the prediction results of each IMF are superimposed to obtain the final predicted wind speed.The results show that the prediction accuracy of the proposed model is better than other typical wind speed prediction models,and the model has good performance in the prediction of ultra-short wind speed.

关 键 词：超短期风速预测 变分模态分解 固有模态分量 去噪 长短期记忆网络 

分 类 号：TM614[电气工程—电力系统及自动化]