为进一步提高超短期电力负荷的预测精度,增强对电力负荷时序特征的提取能力,提出一种考虑相似日与误差修正的时序增强Transformer（TETransformer）超短期电力负荷预测方法。首先,利用灰色关联分析选取气象相似日;然后,在Transformer模型基础上构造局部时序增强注意力机制,利用时序卷积提高注意力机制的局部时序特征感知能力,聚合观测点临近区域相关信息;传统Transformer模型中嵌入时序卷积层,扩展特征图,在Transformer模型全局信息提取的基础上增强局部时序信息提取能力;最后,将历史特征数据和未来气象数据输入TETransforemr,气象相似日的负荷功率序列输入LSTM,通过全连接层融合历史时序特征与相似日信息,引入基于编码器的误差修正模块,提高模型预测精度。通过多模型对比与消融实验,预测精度均有提高,证明所提方法可有效增强对电力负荷的提取能力,在超短期电力负荷领域具有一定的应用意义。
Abstract
In order to further improve the prediction accuracy of ultra-short-term power load and enhance the extraction ability of power load temporal features, a temporal enhancement Transformer （TETransformer） ultra-short-term power load prediction method considering similar days and error correction is proposed. First, the meteorologically similar days are selected using gray relational analysis, and then, the local temporal enhancement attention mechanism is constructed on the basis of the Transformer model, and the temporal convolution is used to improve the local temporal feature perception ability of the attention mechanism, and to aggregate the relevant information of the adjacent area of the observation point; the traditional Transformer model is embedded with a temporal convolution layer, and the feature map is extended, in which the global information of the Transformer model is not available to the public. traditional Transformer model to extend the feature map and enhance the local time-series information extraction capability on the basis of the global information extraction of the Transformer model. Finally, the historical feature data and future meteorological data are input to TETransforemr, and the load power sequences of meteorologically similar days are input to LSTM, and the historical time-series features and similar day information are fused through the fully-connected layer, and the encoder-based error correction module is introduced to improve the model prediction accuracy. Through the multi-model comparison and ablation experiments, the prediction accuracies are improved, which proves that the proposed method can effectively enhance the extraction ability of power load and has certain application significance in the field of ultra-short-term power load.
关键词
负荷功率预测 /
Transformer模型 /
相似日选取 /
灰色关联分析 /
误差修正 /
时序卷积
Key words
load power forecasting /
Transformer model /
similar days selection /
gray relational analysis /
error correction /
temporal convolution
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 韩富佳, 王晓辉, 乔骥, 等. 基于人工智能技术的新型电力系统负荷预测研究综述[J]. 中国电机工程学报, 2023, 43(22): 8569-8592.HAN F J, WANG X H, QIAO J, et al. Review on artificial intelligence based load forecasting research for the new-type power system[J]. Proceedings of the CSEE, 2023, 43(22): 8569-8592.
[2] 陈元峰, 马溪原, 程凯, 等. 基于气象特征量选取与SVM模型参数优化的新能源超短期功率预测[J]. 太阳能学报, 2023, 44(12): 568-576.CHEN Y F, MA X Y, CHENG K, et al. Ultra-short-term power forecast of new energy based on meteorological feature selection and SVM model parameter optimization[J]. Acta energiae solaris sinica, 2023, 44(12): 568-576.
[3] 廖旎焕, 胡智宏, 马莹莹, 等. 电力系统短期负荷预测方法综述[J]. 电力系统保护与控制, 2011, 39(1): 147-152.LIAO N H, HU Z H, MA Y Y, et al. Review of the short-term load forecasting methods of electric power system[J]. Power system protection and control, 2011, 39(1): 147-152.
[4] 方娜, 陈浩, 邓心, 等. 基于VMD-ARIMA-DBN的短期电力负荷预测[J]. 电力系统及其自动化学报, 2023, 35(6): 59-65.FANG N, CHEN H, DENG X, et al. Short-term power load forecasting based on VMD-ARIMA-DBN[J]. Proceedings of the CSU-EPSA, 2023, 35(6): 59-65.
[5] 魏明奎, 叶葳, 沈靖, 等. 基于自组织特征神经网络和最小二乘支持向量机的短期电力负荷预测方法[J]. 现代电力, 2021, 38(1): 17-23.WEI M K, YE W, SHEN J, et al. Short-term load forecasting method based on self-organizing feature mapping neural network and GA-least square SVC model[J]. Modern electric power, 2021, 38(1): 17-23.
[6] 陈振宇, 刘金波, 李晨, 等. 基于LSTM与XGBoost组合模型的超短期电力负荷预测[J]. 电网技术, 2020, 44(2): 614-620.CHEN Z Y, LIU J B, LI C, et al. Ultra-short-term power load forecasting based on LSTM and XGBoost combined model[J]. Power system technology, 2020, 44(2): 614-620.
[7] 李焱, 贾雅君, 李磊, 等. 基于随机森林算法的短期电力负荷预测[J]. 电力系统保护与控制, 2020, 48(21): 117-124.LI Y, JIA Y J, LI L, et al. Short term power load forecasting based on a stochastic forest algorithm[J]. Power system protection and control, 2020, 48(21): 117-124.
[8] 孙辉, 杨帆, 高正男, 等. 考虑特征重要性值波动的MI-BILSTM短期负荷预测[J]. 电力系统自动化, 2022, 46(8): 95-103.SUN H, YANG F, GAO Z N, et al. Short-term load forecasting based on mutual information and bi-directional long short-term memory network considering fluctuation in importance values of features[J]. Automation of electric power systems, 2022, 46(8): 95-103.
[9] 邓斌, 张楠, 王江, 等. 基于LTC-RNN模型的中长期电力负荷预测方法[J]. 天津大学学报(自然科学与工程技术版), 2022, 55(10): 1026-1033.DENG B, ZHANG N, WANG J, et al. Medium-and long-term power load forecasting method based on LTC-RNN model[J]. Journal of Tianjin University (science and technology), 2022, 55(10): 1026-1033.
[10] 朱伟, 孙运全, 钱尧, 等. 基于CEEMD-GRU模型的短期电力负荷预测方法[J]. 电测与仪表, 2023, 60(1): 16-22.ZHU W, SUN Y Q, QIAN Y, et al. Short-term load forecasting method based on complementary ensemble empirical mode decomposition and gated recurrent unit neural network[J]. Electrical measurement & instrumentation, 2023, 60(1): 16-22.
[11] 张未, 余成波, 王士彬, 等. 基于VMD-LSTM-LightGBM的多特征短期电力负荷预测[J]. 南方电网技术, 2023, 17(2): 74-81.ZHANG W, YU C B, WANG S B, et al. Multi-featured short-term power load forecasting based on VMD-LSTM-LightGBM[J]. Southern power system technology, 2023, 17(2): 74-81.
[12] BENGIO Y, SIMARD P, FRASCONI P.Learning long-term dependencies with gradient descent is difficult[J]. IEEE transactions on neural networks, 1994, 5(2): 157-166.
[13] 段雪滢, 李小腾, 陈文洁. 基于改进粒子群优化算法的VMD-GRU短期电力负荷预测[J]. 电工电能新技术, 2022, 41(5): 8-17.DUAN X Y, LI X T, CHEN W J. Improved particles swarm optimization algorithm-based VMD-GRU method for short-term load forecasting[J]. Advanced technology of electrical engineering and energy, 2022, 41(5): 8-17.
[14] 赵兵, 王增平, 纪维佳, 等. 基于注意力机制的CNN-GRU短期电力负荷预测方法[J]. 电网技术, 2019, 43(12): 4370-4376.ZHAO B, WANG Z P, JI W J, et al. A short-term power load forecasting method based on attention mechanism of CNN-GRU[J]. Power system technology, 2019, 43(12): 4370-4376.
[15] VASWANI A, SHAZEER N, PARMAR N, et al.Attention is all you need[J]. Advances in neural information processing systems, 2017, 30: 5998-6008.
[16] ZHOU H Y, ZHANG S H, PENG J Q, et al.Informer: beyond efficient transformer for long sequence time-series forecasting[J]. Proceedings of the AAAI conference on artificial intelligence, 2021, 35(12): 11106-11115.
[17] KITAEV N, KAISER Ł, LEVSKAYA A. Reformer: the efficient transformer[EB/OL]. 2020: arXiv: 2001.04451. https://arxiv.org/abs/2001.04451.
[18] WU H X, XU J H, WANG J M, et al.Autoformer: decomposition transformers with auto-correlation for long-term series forecasting[C]//Neural Information Processing Systems. 2021
[19] WANG L, HE Y G, LI L, et al.A novel approach to ultra-short-term multi-step wind power predictions based on encoder-decoder architecture in natural language processing[J]. Journal of cleaner production, 2022, 354: 131723.
[20] 骆钊, 吴谕侯, 朱家祥, 等. 基于多尺度时间序列块自编码Transformer神经网络模型的风电超短期功率预测[J]. 电网技术, 2023, 47(9): 3527-3537.LUO Z, WU Y H, ZHU J X, et al. Wind power forecasting based on multi-scale time series block auto-encoder Transformer neural network model[J]. Power system technology, 2023, 47(9): 3527-3537.
[21] 王瑞, 闫方, 逯静, 等. 运用相似日和LSTM的短期负荷双向组合预测[J]. 电力系统及其自动化学报, 2022, 34(1): 93-99.WANG R, YAN F, LU J, et al. Bidirectional combined short-term load forecasting by using similar days and LSTM[J]. Proceedings of the CSU-EPSA, 2022, 34(1): 93-99.
[22] 高明, 郝妍. 基于BiLSTM网络与误差修正的超短期负荷预测[J]. 综合智慧能源, 2023, 45(1): 31-40.GAO M, HAO Y. Ultra-short-term load forecasting based on BiLSTM network and error correction[J]. Integrated intelligent energy, 2023, 45(1): 31-40.
[23] 叶家豪, 魏霞, 黄德启, 等. 基于灰色关联分析的BSO-ELM-AdaBoost风电功率短期预测[J]. 太阳能学报, 2022, 43(3): 426-432.YE J H, WEI X, HUANG D Q, et al. Short-term forecast of wind power based on BSO-ELM-AdaBoost with grey correlation analysis[J]. Acta energiae solaris sinica, 2022, 43(3): 426-432.
基金
国家自然科学基金联合基金项目（U21A20482）; 河北建投海上风电有限公司项目（HD2209）
PDF(1365 KB)
