摘要
为提高短期电力负荷预测精度,本文提出了一种基于快照反馈机制改进的变分模态分解技术VMDSF(variational mode decomposition with a snapshot of feedback)和带有循环滑窗策略优化的长短时记忆网络CSLSTM(long short-term memory with circular sliding window)的组合预测方法 VMDSF-CSLSTM。为降低原始序列的不稳定性及复杂性,本文首先使用VMDSF将原始电力负荷序列分解成多个子序列。然后结合网格搜索法对CSLSTM进行最优参数寻找,得到含有最优模型参数的电力负荷短期预测模型。最后,使用2013年澳大利亚4个区域的电力负荷数据集,对本文方法进行算例测试,测试结果表明了本组合模型的有效性。
To improve the accuracy of short-term power load forecasting,a forecasting method which combines varia⁃tional mode decomposition with a snapshot of feedback and long short-term memory with circular sliding window(VMDSF-CSLSTM)is proposed in this paper.To reduce the instability and complexity of the original power load se⁃ries,VMDSF is used to decompose the original series into multiple subsequences at first.Then,the optimal parameters of CSLSTM are searched using the grid search method,thus obtaining the short-term power load forecasting model with the optimal model parameters.Finally,the power load data sets of four regions in Australia in 2013 are taken as a test example,and test results show the effectiveness of the combined model.
作者
孙颢一
易灵芝
刘文翰
邱东洲
赵健
SUN Haoyi;YI Lingzhi;LIU Wenhan;QIU Dongzhou;ZHAO Jian(Hunan Province Engineering Research Center for Multi-Energy Collaborative Control Technology,School of Automation and Electronic Information,Xiangtan University,Xiangtan 411105,China;Hunan Province Cooperative Innovation Center for Wind Power Equipment and Energy Conversion,Xiangtan 411105,China;CHD Tieling Power Generation Co.,Ltd,Tieling 112000,China)
出处
《电力系统及其自动化学报》
CSCD
北大核心
2021年第6期67-74,83,共9页
Proceedings of the CSU-EPSA
关键词
短期负荷预测
快照反馈
循环滑窗策略
变分模态分解
长短时记忆网络
预测精度
short-term load forecasting
snapshot of feedback
circular sliding window strategy
variational mode de⁃composition(VMD)
long short-term memory
forecasting accuracy
