摘要
考虑到风速时间序列非平稳特性和时序关联难以建模的问题,提出一种基于变分模态分解和深度门控循环网络的风速短期预测模型。该模型首先使用变分模态分解非递归地将原始风速序列分解为预先设定层数的子分量,以期降低原始序列的不平稳度,使用深度门控网络分别对各子分量建模预测,最后叠加各分量的预测结果,得到风速的预测结果。实例研究表明所提模型能够有效地跟踪风速的变化,具有较高的短期预测精度。
Considering that it is difficult to model the time dependence and instability of wind speed,a short-term wind speed prediction model based on VMD and deep GRU network is proposed in this paper.In order to reduce the instability of original sequence,VMD is used to obtain the limited number of sub-sequences from the original wind speed sequence.Deep GRU network is adopted to build the prediction model for each sub-sequence.The overall prediction is obtained by superposing each predicting sequence.The case study shows that the proposed model can effectively track the changes of wind speed,and has relatively high accuracy of short-term wind speed prediction.
作者
吴宇杭
温步瀛
江岳文
陈静
王怀远
WU Yuhang;WEN Buying;JIANG Yuewen;CHEN Jing;WANG Huaiyuan(School of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108,Fujian,China)
出处
《电网与清洁能源》
2018年第12期59-64,70,共7页
Power System and Clean Energy
基金
国家自然科学基金项目(51707040)
福建省自然科学基金项目(2018J01482)~~
关键词
风速预测
深度学习
变分模态分解
门控循环网络
wind speed prediction
deep learning
variational mode decomposition
gated recurrent unit
作者简介
吴宇杭(1993—),男,硕士研究生,研究方向为风速及风电功率预测;温步瀛(1967-),男,博士,教授,研究方向为电力系统优化运行与电力市场;江岳文(1977-),女,博士,教授,研究方向为风电并网优化运行与电力市场。
