周期规律增强的多视角短期电力负荷预测

苏伟; 肖小龙; 史明明; 方鑫; 司鑫尧

doi:10.13700/j.bh.1001-5965.2022.0399

周期规律增强的多视角短期电力负荷预测

doi: 10.13700/j.bh.1001-5965.2022.0399

国网江苏省电力有限公司电力科学研究院，南京 210000

基金项目: 国网江苏省电力有限公司科技项目(J2021165)

详细信息

通讯作者:
E-mail： susw2011@163.com

中图分类号: TP399
计量
- 文章访问数: 179
- HTML全文浏览量: 58
- PDF下载量: 4
- 被引次数: 0
出版历程
- 收稿日期: 2022-05-21
- 录用日期: 2022-07-02
- 网络出版日期: 2022-12-14
- 整期出版日期: 2024-02-27

Periodic pattern-enhanced multi-view short-term load prediction

State Grid Jiangsu Electric Power Co.，Ltd. Research Institute，Nanjing 210000，China

Funds: The Science and Technology Project of State Grid Jiangsu Electric Power Co., Ltd (J2021165)

More Information

Corresponding author: E-mail: susw2011@163.com

摘要

摘要:
短期电力负荷预测对电力系统的可靠运行具有重要意义。现有方法存在如下问题：缺乏对特征之间依赖关系的挖掘；忽略了电力负荷变化的周期性规律。为此，提出一种周期规律增强的多视角短期电力负荷预测网络（EPISODE）方法。EPISODE方法主要包括2个核心组件：多视角特征学习组件和周期规律增强的电力负荷预测组件。前者旨在有效提取电力负荷数据中的静态特征与时序特征，以得到增强的特征表示；后者则是对电力负荷数据进行一般性时序挖掘和周期性时序挖掘，从而得到全面的电力负荷历史数据表征。基于后期融合的方式，实现短期电力负荷预测。在真实公开的电力负荷预测数据集上进行了大量实验。实验结果证明了所提方法相比现有基准方法的先进性。
- 周期规律增强 /
- 短期负荷预测 /
- 特征压缩与激励 /
- 门控循环单元 /
- 注意力神经网络
Abstract:
Short-term load prediction is essential to ensure the proper operation of the power system. The existing efforts have two limitations: lack of mining the dependencies between features and ignore the periodic pattern of load changes. To solve the above limitations, we propose periodic Pattern-enhanced MultI-view Short-term power load prediction nEtworks, dubbed EPISODE. The framework includes two core components: a multi-view feature learning component and a periodic pattern-enhanced load prediction component. The former aims to effectively extract static features and time series features to obtain enhanced feature representation; the latter is to perform general time series mining and periodic time series mining to obtain comprehensive historical feature representation. The combination of the two aforementioned qualities results in the realization of the short-term load forecast. Extensive experiments have been conducted on real-world datasets, and the experimental results demonstrate the superiority of our proposed method.
- periodic pattern-enhanced /
- short-term load prediction /
- feature squeeze and excitation /
- gated recurrent unit /
- attention mechanism

HTML全文

图 1 EPISODE整体网络结构图

Figure 1. Overall network structure diagram of EPISODE

下载: 全尺寸图片幻灯片

图 2 基于一维卷积神经网络的静态特征提取模块

Figure 2. 1D-CNN based static feature extraction module

下载: 全尺寸图片幻灯片

图 3 压缩激励模块结构

Figure 3. Squeeze-and-excitation module structure

下载: 全尺寸图片幻灯片

图 4 基于GRU的时序特征提取模块结构

Figure 4. GRU based time series feature extraction module structure

下载: 全尺寸图片幻灯片

图 5 EPISODE在不同时间窗口下的预测结果

Figure 5. Prediction results of EPISODE in different windows

下载: 全尺寸图片幻灯片

图 6 EPISODE与基准方法的预测结果

Figure 6. The load prediction results of EPISODE and baselines

下载: 全尺寸图片幻灯片

图 7 EPISODE 与衍生模型的预测结果

Figure 7. The load prediction results of EPISODE and derivations

下载: 全尺寸图片幻灯片

图 8 特征提取单元层数的敏感性实验结果

Figure 8. Sensitivity experiment results with different feature extraction layers

下载: 全尺寸图片幻灯片

图 9 GRU-Skip模块中$ p $取值的敏感性实验结果

Figure 9. Sensitivity experiments with different value of $ p $ in GRU-Skip module

下载: 全尺寸图片幻灯片

表 1 EPISODE在不同时间窗口的预测精度对比

Table 1. Comparison of prediction accuracy of EPISODE in different window

时间窗口	MAPE/%	RMSE/MW
$ M= $7	1.5234	165.7249
$ M=14 $	1.3993	112.4938
$ M=21 $	1.0115	81.2460
$ M=28 $	1.3053	115.5637
$ M=30 $	1.1400	100.8728

下载: 导出CSV

表 2 EPISODE与基准方法在数据集上的预测精度和测试时间对比

Table 2. The comparison of EPISODE and baselines methods on prediction accuracy and test time

方法	MAPE/%	RMSE/MW	时间/s
EPISODE	1.0115	81.2460	0.2953
Attention-LSTM	2.7877	233.0022	0.7178
FOMNN	2.1967	194.7047	0.3050
GRU-NN	6.2449	472.8429	0.7624
Dropout-ILSTM	5.5497	392.0774	0.9222
Attention-BiLSTM-LSTM	2.1984	194.7577	1.5094
CNN-BiLSTM	4.1792	422.5311	2.7181

下载: 导出CSV

表 3 EPISODE与衍生模型的预测精度对比

Table 3. Comparison of prediction accuracy between EPISODE and derivations

模型	MAPE/%	RMSE/MW
EPISODE	1.0115	81.2460
EPISODE_ NoSE	1.4542	123.2089
EPISODE_NoGRU_Skip	1.6223	168.6200
EPISODE_NoAttention	1.6427	188.6485

下载: 导出CSV

参考文献(25)

[1]	康重庆, 夏清, 张伯明. 电力系统负荷预测研究综述与发展方向的探讨[J]. 电力系统自动化, 2004, 28(17): 1-11. KANG C Q, XIA Q, ZHANG B M. Review of power system load forecasting and its development[J]. Automation of Electric Power Systems, 2004, 28(17): 1-11 (in Chinese).
[2]	KEMP F. Applied multiple regression/correlation analysis for the behavioral sciences[J]. Journal of the Royal Statistical Society: Series D (the Statistician), 2003, 52(4): 691-691.
[3]	DI CAPRIO U, GENESIO R, POZZI S, et al. Short term load forecasting in electric power systems: A comparison of ARMA models and extended Wiener filtering[J]. Journal of Forecasting, 1983, 2(1): 59-76. doi: 10.1002/for.3980020107
[4]	LEE Y S, TONG L I. Forecasting time series using a methodology based on autoregressive integrated moving average and genetic programming[J]. Knowledge-Based Systems, 2011, 24(1): 66-72. doi: 10.1016/j.knosys.2010.07.006
[5]	H ERNÁNDEZ L, BALADRÓN C, AGUIAR J, et al. Artificial neural network for short-term load forecasting in distribution systems[J]. Energies, 2014, 7(3): 1576-1598. doi: 10.3390/en7031576
[6]	LDIN W M, TU C S, YANG R F, et al. Particle swarm optimisation aided least-square support vector machine for load forecast with spikes[J]. IET Generation, Transmission & Distribution, 2016, 10(5): 1145-1153.
[7]	吴倩红, 高军, 侯广松, 等. 实现影响因素多源异构融合的短期负荷预测支持向量机算法[J]. 电力系统自动化2016, 40(15): 67-72. WU Q H, GAO J, HOU G S, et al. Short-term load forecasting support vector machine algorithm based on multi-source heterogeneous fusion of load factors[J]. Automation of Electric Power System, 2016, 40(15): 67-72 (in Chinese)
[8]	李婉华, 陈宏, 郭昆, 等. 基于随机森林算法的用电负荷预测研究[J]. 计算机工程与应用, 2016, 52(23): 236-243. LI W H, CHEN H, GUO K, et al. Research on electrical load prediction based on random forest algorithm[J]. Computer Engineering and Applications, 2016, 52(23): 236-243(in Chinese).
[9]	彭鹏, 彭佳红. 基于多元线性回归模型的电力负荷预测研究[J]. 中国安全生产科学技术, 2011, 7(9): 158-161. PENG P, PENG J H. Research on the prediction of power load based on multiple linear regression model[J]. Journal of Safety Science and Technology, 2011, 7(9): 158-161 (in Chinese).
[10]	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Imagenet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90.
[11]	RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088), 533-536.
[12]	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. doi: 10.1162/neco.1997.9.8.1735
[13]	CHUNG J, GULCEHRE C, CHO K, et al. Gated feedback recurrent neural networks[C]//International Conference on Machine Learning. Lile: PMLR, 2015: 2067-2075.
[14]	彭文, 王金睿, 尹山青. 电力市场中基于Attention-LSTM的短期负荷预测模型[J]. 电网技术, 2019, 43(5): 1745-1751. PENG W, WANG J R, YIN S Q. Short-term load forecasting model based on attention-LSTM in electricity market[J]. Power System Technology, 2019, 43(5): 1745-1751 (in Chinese).
[15]	李龙, 魏靖, 黎灿兵, 等. 基于人工神经网络的负荷模型预测[J]. 电工技术学报, 2015, 30(8): 225-230. LI L, WEI J, LI C B, et al. Prediction of load model based on artificial neural network[J]. Transactions of China Electrotechnical Society, 2015, 30(8): 225-230 (in Chinese).
[16]	陆继翔, 张琪培, 杨志宏, 等. 基于CNN-LSTM混合神经网络模型的短期负荷预测方法[J]. 电力系统自动化, 2019, 43(8): 131-137. LU J X, ZHANG Q P, YANG Z H, et al. Short-term load forecasting method based on CNN-LSTM hybrid neural network model[J]. Automation of Electric Power Systems, 2019, 43(8): 131-137 (in Chinese).
[17]	李泽文, 胡让, 刘湘, 等. 基于PCA-DBILSTM的多因素短期负荷预测模型[J]. 电力系统及其自动化学报, 2020, 32(12): 32-39. LI Z W, HU R, LIU X, et al. Multi-factor short-term load prediction model based on PCA-DBILSTM[J]. Proceedings of the CSU-EPSA, 2020, 32(12): 32-39 (in Chinese).
[18]	杨洋, 谷震浩. 基于XGBOOST-DNN的中期电力负荷预测[J]. 计算机系统应用, 2021, 30(9): 186-191. doi: 10.15888/j.cnki.csa.008098 YANG Y, GU Z H. Mid-term power load forecasting based on XGBoost-DNN[J]. Computer Systems & Applications, 2021, 30(9): 186-191 (in Chinese). doi: 10.15888/j.cnki.csa.008098
[19]	HU J, SHEN L, ALBANIE S, et al. Squeeze-and-excitation networks[EB/OL]. (2019-05-16) [2022-05-18]. https://arxiv.org/abs/1709.01507.pdf.
[20]	LAI G K, CHANG W C, YANG Y M, et al. Modeling long- and short-term temporal patterns with deep neural networks[EB/OL]. (2018-04-18) [2022-05-18] .https://arxiv.org/abs/1703.07015.pdf.
[21]	庞昊, 高金峰, 杜耀恒. 基于多神经网络融合的短期负荷预测方法[J]. 电力自动化设备, 2020, 40(6): 37-43. PANG H, GAO J F, DU Y H. Short-term load forecasting method based on fusion of multiple neural networks[J]. Electric Power Automation Equipment, 2020, 40(6): 37-43 (in Chinese).
[22]	王增平, 赵兵, 纪维佳, 等. 基于GRU-NN模型的短期负荷预测方法[J]. 电力系统自动化, 2019, 43(5): 53-58. WANG Z P, ZHAO B, JI W J, et al. Short-term load forecasting method based on GRU-NN model[J]. Automation of Electric Power Systems, 2019, 43(5): 53-58 (in Chinese).
[23]	刘皓琪, 高飞, 王耀力, 等. 基于Dropout-ILSTM网络的短期电力负荷预测[J]. 电测与仪表, 2021, 58(5): 105-111. LIU H Q, GAO F, WANG Y L, et al. Short-term power load forecasting based on Dropout-ILSTM network[J]. Electrical Measurement & Instrumentation, 2021, 58(5): 105-111 (in Chinese).
[24]	龚飘怡, 罗云峰, 方哲梅, 等. 基于Attention-BiLSTM-LSTM神经网络的短期电力负荷预测方法[J]. 计算机应用, 2021, 41(S1): 81-86. GONG P Y, LUO Y F, FANG Z M, et al. Short-term power load forecasting method based on Attention-BiLSTM-LSTM neural network[J]. Journal of Computer Applications, 2021, 41(S1): 81-86 (in Chinese).
[25]	朱凌建, 荀子涵, 王裕鑫, 等. 基于CNN-Bi LSTM的短期电力负荷预测[J]. 电网技术, 2021, 45(11): 4532-4539. ZHU L J, XUN Z H, WANG Y X, et al. Short-term power load forecasting based on CNN-BiLSTM[J]. Power System Technology, 2021, 45(11): 4532-4539 (in Chinese).