[1]张蕾,季媛媛,李娜,等.基于TCN与双重注意力机制的光伏功率预测模型[J].计算机技术与发展,2025,(07):214-220.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0045]
　ZHANG Lei,JI Yuan-yuan,LI Na,et al.A Photovoltaic Power Prediction Model Based on TCN and Double Attention Mechanism[J].,2025,(07):214-220.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0045]
点击复制

基于TCN与双重注意力机制的光伏功率预测模型

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年07期
页码:
214-220
栏目:
新型计算应用系统
出版日期:
2025-07-10

文章信息/Info

Title:
A Photovoltaic Power Prediction Model Based on TCN and Double Attention Mechanism
文章编号:
1673-629X(2025)07-0214-07
作者:
张蕾1; 2; 季媛媛1; 2; 李娜1; 2; 朱俊澎3
1. 国网江苏省电力有限公司南通供电分公司,江苏 南通 226000; 2. 南通电力设计院有限公司,江苏 南通 226000; 3. 河海大学 电气与动力工程学院,江苏 南京 210024
Author(s):
ZHANG Lei1; 2; JI Yuan-yuan1; 2; LI Na1; 2; ZHU Jun-peng3
1. State Grid Jiangsu Nantong Electric Power Co. ,Ltd. ,Nantong 226000,China; 2. Nantong Electric Power Design Institute Co. ,Ltd. ,Nantong 226000,China; 3. School of Electrical and Power Engineering,Hohai University,Nanjing 210024,China
关键词:
深度学习; 光伏功率预测; 时域卷积网络; 多头注意力机制; 外部注意力机制
Keywords:
deep learning;  photovoltaic power prediction;  temporal convolutional network; multi - head attention mechanism;  external attention mechanism
分类号:
TP183;TM615
DOI:
10.20165/j.cnki.ISSN1673-629X.2025.0045
摘要:
针对传统光伏输出功率预测模型准确度不高、训练效率较低的问题,提出一种基于时域卷积网络(TCN)与双重注意力机制(DAM)的光伏功率预测模型。 首先,利用时域卷积网络中的第一个因果扩张卷积层对输入序列数据进行因果扩张卷积运算,初步提取数据短期依赖关系;随后,采用多头注意力机制对重要时间步进行加权,从而捕捉序列数据的长期依赖关系;接着,通过时域卷积网络的第二个因果扩张卷积层进一步提取数据深层次特征,以增强模型对数据短期依赖特征的提取能力;然后,利用外部注意力机制将外部信息及第二个因果扩张卷积层的输出映射到隐藏特征空间,并通过非线性函数生成注意力权重分布,实现特征相关性权重的动态分配;最后,结合残差连接实现对光伏功率的高效预测。 算例分析结果表明,在实际测试中该模型的平均绝对误差、均方根误差和绝对系数分别为 473. 58 kW、828. 77 kW 和 98. 27% ,相较于传统光伏功率预测模型具有更高的预测精度,且模型训练耗时短,计算效率优良。
Abstract:
To address the issues of low accuracy and inefficient training in traditional photovoltaic (PV) power prediction models,a pho-tovoltaic power prediction model based on Temporal Convolutional Network ( TCN) and Double Attention Mechanism ( DAM) is proposed. First,the initial causal dilated convolution operation is performed on the input sequence data through the first causal dilated convolution layer of the TCN to preliminarily extract short-term dependencies. Next,the multi-head attention mechanism is applied to weight important time steps,capturing the long-term dependencies in the sequence data. Then,the second causal dilated convolution layer of the TCN further extracts deep-level features,enhancing the model's ability to capture short-term dependencies. Afterward,an external attention mechanism is used to map external information and the output of the second causal dilated convolution layer into a hidden feature space,where attention weight distributions are generated through a nonlinear function to dynamically allocate feature relevance weights. Finally,residual connections are integrated to achieve efficient PV power prediction. The results of the case study analysis show that the proposed model in the actual test set achieved an average absolute error of 473. 58 kW,a root mean square error of 828. 77 kW,and an absolute coefficient of 98. 27% . Compared to traditional photovoltaic power forecasting models,it demonstrates higher prediction accuracy,shorter training time,and excellent computational efficiency.

相似文献/References:

[1]陈强锐,谢世朋.基于深度学习的肺部肿瘤检测方法[J].计算机技术与发展,2018,28(04):201.[doi:10.3969/ j. issn.1673-629X.2018.04.043]
　CHEN Qiang-rui,XIE Shi-peng.Lung Cancer Detection Method Based on Deep Learning[J].,2018,28(07):201.[doi:10.3969/ j. issn.1673-629X.2018.04.043]
[2]施泽浩,赵启军.基于全卷积网络的目标检测算法[J].计算机技术与发展,2018,28(05):55.[doi:10.3969/j.issn.1673－629X.2018.05.013]
　SHI Ze-hao,ZHAO Qi-jun.Object Detection Algorithm Based on Fully Convolutional Neural Network[J].,2018,28(07):55.[doi:10.3969/j.issn.1673－629X.2018.05.013]
[3]黄法秀,张世杰,吴志红,等.数据增广下的人脸识别研究[J].计算机技术与发展,2020,30(03):67.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 013]
　HUANG Fa-xiu,ZHANG Shi-jie,WU Zhi-hong,et al.Research on Face Recognition Based on Data Augmentation[J].,2020,30(07):67.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 013]
[4]陈浩翔,蔡建明,刘铿然,等. 手写数字深度特征学习与识别[J].计算机技术与发展,2016,26(07):19.
　CHEN Hao-xiang,CAI Jian-ming,LIU Keng-ran,et al. Deep Learning and Recognition of Handwritten Numeral Features[J].,2016,26(07):19.
[5]高翔,陈志,岳文静,等.基于视频场景深度学习的人物语义识别模型[J].计算机技术与发展,2018,28(06):53.[doi:10.3969/ j. issn.1673-629X.2018.06.012]
　GAO Xiang,CHEN Zhi,YUE Wen-jing,et al.Human Semantic Recognition Model Based on Video Scene Deep Learning[J].,2018,28(07):53.[doi:10.3969/ j. issn.1673-629X.2018.06.012]
[6]贺飞翔,赵启军. 基于深度学习的头部姿态估计[J].计算机技术与发展,2016,26(11):1.
　HE Fei-xiang,ZHAO Qi-jun. Head Pose Estimation Based on Deep Learning[J].,2016,26(07):1.
[7]徐 融,邱晓晖.一种改进的 YOLO V3 目标检测方法[J].计算机技术与发展,2020,30(07):30.[doi:10. 3969 / j. issn. 1673-629X. 2020. 07. 007]
　XU Rong,QIU Xiao-hui.An Improved YOLO V3 Object Detection[J].,2020,30(07):30.[doi:10. 3969 / j. issn. 1673-629X. 2020. 07. 007]
[8]曾志平[] [],萧海东[],张新鹏[]. 基于DBN的金融时序数据建模与决策[J].计算机技术与发展,2017,27(04):1.
　ZENG Zhi-ping[] [],XIAO Hai-dong[],ZHANG Xin-peng[]. Modeling and Decision-making of Financial Time Series Data with DBN[J].,2017,27(07):1.
[9]李全兵,文 钊*,田艳梅*,等.基于 WGAN 的音频关键词识别研究[J].计算机技术与发展,2021,31(08):26.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 005]
　LI Quan-bing,WEN Zhao *,TIAN Yan-mei *,et al.Research on Audio Keywords Recognition Based on WassersteinGenerative Adversarial Network[J].,2021,31(07):26.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 005]
[10]李宏林. 分析式纹理合成技术及其在深度学习的应用[J].计算机技术与发展,2017,27(11):7.
　LI Hong-lin. Analyzed Texture-synthesis Techniques and Their Applications in Deep Learning[J].,2017,27(07):7.
[11]陈君,郭立颖*,赵小会,等.基于MPBiLSTM的短期光伏发电功率预测[J].计算机技术与发展,2024,34(10):186.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0204]
　CHEN Jun,GUO Li-ying*,ZHAO Xiao-hui,et al.Short-term Photovoltaic Power Prediction Based on MPBiLSTM[J].,2024,34(07):186.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0204]
[12]周新宇,姜志航,白峻铭,等.基于IDBO-TCN-LSTM的短期光伏功率预测[J].计算机技术与发展,2025,(07):125.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0044]
　ZHOU Xin-yu,JIANG Zhi-hang,BAI Jun-ming,et al.Short-term PV Power Prediction Based on IDBO-TCN-LSTM[J].,2025,(07):125.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0044]

更新日期/Last Update: 2025-07-10