摘要
短期电力负荷预测能够评估某地区整体的电力负荷变化情况,对电力系统安全稳定运行至关重要,而气象因素也具有深刻的影响。作者提出一种融合了经验模态分解(EMD)和双向长短期记忆网络(BiLSTM)的模型,首先结合气象因素分析其与电力系统用电量的相关性,选取有相关性的气象因素和历史负荷数据一起作为输入特征集,利用EMD算法将随机性强的历史电力负荷数据分解为有限个特征互异的固有模态函数分量和趋势分量。然后和气象因素一起输入到BiLSTM中深度挖掘历史数据并训练模型。最后对各分量数据分别预测并叠加输出预测值。以某地电力负荷数据为实际算例,算例结果表明,采用该方法预测模型拟合度能达到97%,具有较好的预测效果。相较于LSTM网络和BiLSTM网络的预测结果,其预测曲线更贴近于历史负荷数据,特别是对于电力负荷趋势的突然变化,其预测精度得到有效提升。
Short term power load forecasting can evaluate the overall power load changes in a certain region,which is crucial for the safe and stable operation of the power system,and meteorological factors also have a profound impact.This article proposes a model that combines Empirical Mode Decomposition(EMD)and Bidirectional Long Short Term Memory Network(BiLSTM).Firstly,the correlation between meteorological factors and power system consumption is analyzed,and relevant meteorological factors and historical load data are selected as input feature sets.The EMD algorithm is used to decompose the historical power load data with strong randomness into a finite number of intrinsic mode function components and trend components with distinct features.Then,along with meteorological factors,it is input into BiLSTM to deeply mine historical data and train the model.Finally,predict and overlay the predicted values for each component data separately.Taking the power load data of a certain area as an actual calculation example,the experimental results show that the fitting degree of the prediction model using this method can reach 97%,and it has a good prediction effect.Compared to the prediction results of LSTM network and BiLSTM network,its prediction curve is closer to historical load data,especially for sudden changes in power load trends,which greatly improves its prediction accuracy.
作者
沈旃葳
吴细秀
罗格帅
秦理
张清勇
SHEN Zhan-wei;WU Xi-xiu;LUO Ge-shuai;QIN Li;ZHANG Qing-yong(School of Automation,Wuhan University of Technology,Wuhan 430070,China;School of Transportation and Logistics,Wuhan University of Technology,Wuhan 430063,China;Guangdong Power Grid Energy Development Co Ltd,Guangzhou 510160,China)
出处
《武汉理工大学学报》
CAS
2024年第2期141-149,共9页
Journal of Wuhan University of Technology
关键词
短期电力负荷预测
经验模态分解
双向长短期记忆网络
时间序列
short term power load forecasting
empirical mode decomposition
bidirectional long-term and shortterm memory network
time series
作者简介
沈旃葳(1998-),女,硕士生.E-mail:893060664@qq.com。
