摘 要:针对电价非平稳性强的问题,为提高电价的预测精度,提出一种基于变分模态分解VMD)、经验模式分解(EMD)和长短时记忆网络(LSTM)的混合预测模型。首先利用 VMD 将电价分解为若干子序列和残差项;针对残差项具有较强非平稳性的问题,利用 EMD 对残差项进一步分解;最后对各子序列分别利用 LSTM 模型进行预测,并将各子序列预测结果叠加得到最终预测电价。实验结果表明,该方法相比于其他对比方法具有更高的预测精度。
关键词:电价预测;变分模态分解;经验模式分解;长短时记忆网络
DOI:10.19850/j.cnki.2096-4706.2022.18.020
中图分类号:TM715 文献标识码:A 文章编号:2096-4706(2022)18-0084-05
Day-Ahead Electricity Price Forecast Based on VMD-EMD-LSTM
ZHAI Guangsong, WANG Peng, XIE Zhifeng, WU Zhenbo
(School of Automation, Guangdong University of Technology, Guangzhou 510006, China)
Abstract: Aiming at the problem of strong un-stationarity of electricity price, in order to improve the forecast accuracy of electricity price, a hybrid forecast model based on Variational Mode Decomposition (VMD), Empirical Mode Decomposition (EMD) and Long Short-Term Memory Network (LSTM) is proposed. Firstly, it uses VMD to decompose the electricity price into several sub-sequences and residual terms. Then, for the problem that the residual terms have strong un-stationarity, it uses EMD to further decompose the residual terms. Finally, it uses LSTM model to forecast each sub-sequence respectively, and superimposes the forecast results of each subsequence to obtain the final forecast electricity price. The experimental results show that this method has higher forecast accuracy than other contrasting methods.
Keywords: electricity price forecast; Variational Mode Decomposition; Empirical Mode Decomposition; Long Short-Term Memory Network
参考文献:
[1] JESUS L,GRZEGORZ M,BART D S,et al. Forecasting day-ahead electricity prices:a review of state-of-the-art algorithms,best practices and an open-access benchmark [J].Applied Energy,2021,293.
[2] 殷豪,曾云,孟安波,等 . 基于奇异谱分析的短期电价预测 [J]. 电力系统保护与控制,2019,47(1):115-122.
[3] 魏勤,陈仕军,黄炜斌,等 . 利用随机森林回归的现货市场出清价格预测方法 [J]. 中国电机工程学报,2021,41(4):1360-1367+1542.
[4] 姚子麟,张亮,邹斌,等 . 含高比例风电的电力市场电价预测 [J]. 电力系统自动化,2020,44(12):49-55.
[5] 刘达,雷自强,孙堃 . 基于小波包分解和长短期记忆网络的短期电价预测 [J]. 智慧电力,2020,48(4):77-83.
[6] 勾玄,肖先勇 . 基于经验模式分解与 LSTM 神经网络的短期电价预测模型 [J]. 西安理工大学学报,2020,36(1):129-134.
[7] 杨昭,张钢,赵俊杰,等 . 基于变分模态分解和改进粒子群算法优化最小二乘支持向量机的短期电价预测 [J]. 电气技术,2021,22(10):11-16.
作者简介:翟广松(1997—),男,汉族,河南新乡人,硕士研究生在读,研究方向:人工智能算法在电力系统中的应用;王鹏(1998—),男,汉族,宁夏吴忠人,硕士研究生在读,研究方向:人工智能算法在电力系统中的应用;谢智锋(1998—),男,汉族,广东广州人,硕士研究生在读,研究方向:人工智能算法在电力系统中的应用。