唐雅洁,龚迪阳,林达,陈乐祺,邵方格
TANG Yajie,GONG Diyang,LIN Da,CHEN Leqi,SHAO Fangge

• 1:国网浙江省电力有限公司电力科学研究院

摘要(Abstract)：

提高分布式光伏超短期功率预测的准确性对于优化电网实时调度、保障供电可靠性及制定电力市场竞争策略具有重要意义。因此，提出一种基于SAX（符号聚合近似）-MHA（多头注意力）-GCN（图卷积网络）的区域分布式光伏超短期功率预测方法。首先，选取近时段分布式光伏出力时序数据作为输入特征集，并利用SAX算法对其进行压缩，构建图卷积特征矩阵与邻接矩阵。接着，构建深度融合的MHA-GCN算法框架，在每个图卷积块中，先通过时空多头注意力机制层提取特征间的依赖关系，再进行空间图卷积与时间卷积运算，经残差连接后得到模块输出。然后，通过全连接层以及SAX的反向过程，得到超短期功率预测值。最后，以浙江某区域分布式光伏集群为实验对象进行算例分析，结果表明，所提方法在区域整体和单场站预测效果均优于传统方法，有效提升了预测精度与预测时效性。
Improving the accuracy of ultra-short-term power forecasting for distributed photovoltaic(DPV) systems is of great significance for optimizing real-time grid dispatch, ensuring power supply reliability, and formulating electricity market competition strategies. Therefore, this paper proposes an ultra-short-term power forecasting method for regional DPV based on symbolic aggregate approximation(SAX), multi-head attention(MHA), and graph convolutional network(GCN). Firstly, the recent time-series data of DPV system output is selected as the input feature set, and the SAX method is utilized to compress the data, constructing the feature matrix and adjacency matrix of the GCN. Subsequently, a deeply integrated MHA-GCN framework is built. In each graph convolution block, the spatio-temporal MHA mechanism layer is first employed to extract the dependencies between features, followed by spatial graph convolution and temporal convolution operations. The module output is then obtained through residual connections. Afterwards, the ultra-short-term power forecasting values are derived via a fully connected layer and the inverse process of the SAX. Finally, case studies are conducted on a DPV system cluster in a certain region of Zhejiang Province. The results demonstrate that the proposed method outperforms traditional methods in both regional and single-station forecasting, effectively enhancing forecasting accuracy and timeliness.

关键词(KeyWords)： 区域分布式光伏;超短期功率预测;符号聚合近似;多头注意力机制;时空图卷积网络
DPV system;ultra-short-term power forecasting;SAX;MHA mechanism;spatio-temporal GCN

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金联合基金重点项目(U24B2080);; 国网浙江省电力有限公司科技项目(5211DS230003)

作者(Author): 唐雅洁,龚迪阳,林达,陈乐祺,邵方格
TANG Yajie,GONG Diyang,LIN Da,CHEN Leqi,SHAO Fangge

DOI: 10.19585/j.zjdl.202508009

参考文献(References)：

• [1]黎博，陈民铀，钟海旺，等.高比例可再生能源新型电力系统长期规划综述[J].中国电机工程学报，2023,43(2):555-581.LI Bo,CHEN Minyou,ZHONG Haiwang,et al.A review of long-term planning of new power systems with large share of renewable energy[J].Proceedings of the CSEE,2023,43(2):555-581.
• [2]韩肖清，李廷钧，张东霞，等.双碳目标下的新型电力系统规划新问题及关键技术[J].高电压技术，2021,47(9):3036-3046.HAN Xiaoqing,LI Tingjun,ZHANG Dongxia,et al.New issues and key technologies of new power system planning under double carbon goals[J].High Voltage Engineering,2021,47(9):3036-3046.
• [3]栗峰，丁杰，周才期，等.新型电力系统下分布式光伏规模化并网运行关键技术探讨[J].电网技术，2024,48(1):184-199.LI Feng,DING Jie,ZHOU Caiqi,et al.Key technologies of large-scale grid-connected operation of distributed photovoltaic under new-type power system[J].Power System Technology,2024,48(1):184-199.
• [4]关立，黄国栋，吴锋，等.面向高比例分布式光伏的电力调度及市场化机制研究[J].浙江电力，2022,41(8):10-16.GUAN Li,HUANG Guodong,WU Feng,et al.Research on power dispatching and marketization mechanism for high-proportion distributed photovoltaic[J].Zhejiang Electric Power,2022,41(8):10-16.
• [5]国家市场监督管理总局，国家标准化管理委员会.调度侧风电或光伏功率预测系统技术要求：GB/T 40607—2021[S].北京：中国标准出版社，2021.
• [6]孙玲玲，李海滨，贾清泉，等.面向用户侧分布式光伏资源集群开发的博弈策略[J].电力系统自动化，2023,47(10):26-37.SUN Lingling,LI Haibin,JIA Qingquan,et al.Game strategy for cluster development of user-side distributed photovoltaic resources[J]. Automation of Electric Power Systems,2023,47(10):26-37.
• [7]赖昌伟，黎静华，陈博，等.光伏发电出力预测技术研究综述[J].电工技术学报，2019,34(6):1201-1217.LAI Changwei,LI Jinghua,CHEN Bo,et al. Review of photovoltaic power output prediction technology[J].Transactions of China Electrotechnical Society,2019,34(6):1201-1217.
• [8]董存，王铮，白捷予，等.光伏发电功率超短期预测方法综述[J].高电压技术，2023,49(7):2938-2951.DONG Cun,WANG Zheng,BAI Jieyu,et al.Review of ultra-short-term forecasting methods for photovoltaic power generation[J].High Voltage Engineering,2023,49(7):2938-2951.
• [9]臧海祥，程礼临，刘玲，等.基于数据驱动的太阳辐射估计和预测研究与展望[J].电力系统自动化，2021,45(11):170-183.ZANG Haixiang,CHENG Lilin,LIU Ling,et al.Research and prospect for data-driven estimation and prediction of solar radiation[J].Automation of Electric Power Systems,2021,45(11):170-183.
• [10]黄泽，毕贵红，谢旭，等.基于MBI-PBI-ResNet的超短期光伏功率预测[J].电力系统保护与控制，2024,52(2):165-176.HUANG Ze,BI Guihong,XIE Xu,et al.Ultra-short-term PV power prediction based on MBI-PBI-res net[J].Power System Protection and Control,2024,52(2):165-176.
• [11]杨鹏伟，赵丽萍，陈军法，等.基于相似时段匹配与Transformer网络建模的分布式光伏超短期功率预测方法[J].中国电力，2024,57(12):60-70.YANG Pengwei,ZHAO Liping,CHEN Junfa,et al.Distributed photovoltaic ultra-short-term power forecasting method based on temporal analog matching approach and transformer network modeling[J]. Electric Power,2024,57(12):60-70.
• [12]郑珂，王丽婕，郝颖，等.基于数据集蒸馏的光伏发电功率超短期预测[J].中国电机工程学报，2024,44(13):5196-5208.ZHENG Ke,WANG Lijie,HAO Ying,et al.Ultra-shortterm prediction of photovoltaic power based on dataset distillation[J].Proceedings of the CSEE,2024,44(13):5196-5208.
• [13]阮呈隆，李康平，李正辉，等.分布式光伏集群功率短期预测的空间互补特性初探[J].电力系统自动化，2024,48(3):42-50.RUAN Chenglong,LI Kangping,LI Zhenghui,et al.Preliminary study on spatial complementarity characteristics of short-term power prediction for distributed photovoltaic clusters[J].Automation of Electric Power Systems,2024,48(3):42-50.
• [14]唐雅洁，龚迪阳，倪筹帷，等.基于邻域前向时序最优组合的分布式光伏超短期功率预测[J].浙江电力，2021,40(10):95-101.TANG Yajie,GONG Diyang,NI Chouwei,et al. Ultra short-term distributed photovoltaic power forecasting based on neighboring optimal forward sequential combination[J].Zhejiang Electric Power,2021,40(10):95-101.
• [15]严华江，章坚民，胡瑛俊，等.考虑空间相关性的分布式光伏发电出力预测及误差评价指标研究[J].浙江电力，2020,39(3):54-60.YAN Huajiang,ZHANG Jianmin,HU Yingjun,et al.Distributed photovoltaic power generation output prediction based on spatial correlation and error evaluation indexes[J].Zhejiang Electric Power,2020,39(3):54-60.
• [16] LIN J,KEOGH E,LONARDI S,et al.A symbolic representation of time series, with implications for streaming algorithms[C]//Proceedings of the 8th ACM SIGMOD Workshop on Research Issues in Data Mining and Knowledge Discovery.13 June 2003,San Diego,California.ACM,2003:2-11.
• [17]周赣，茅欢，冯燕钧，等.基于多特征符号聚合近似和层次聚类的户变关系识别方法[J].电力系统自动化，2024,48(3):133-141.ZHOU Gan,MAO Huan,FENG Yanjun,et al.Identification method for user-transformer relationship based on multi-feature symbolic aggregate approximation and hierarchical clustering[J]. Automation of Electric Power Systems,2024,48(3):133-141.
• [18] GUO S N,LIN Y F,FENG N,et al. Attention based spatial-temporal graph convolutional networks for traffic flow forecasting[C]//Proceedings of the ThirtyThird AAAI Conference on Artificial Intelligence and Thirty-First Innovative Applications of Artificial Intelligence Conference and Ninth AAAI Symposium on Educational Advances in Artificial Intelligence. ACM,2019:922-929.
• [19]李立刚，孙立群，戴永寿，等.基于时空图卷积神经网络的光伏发电功率超短期预测方法[J].热能动力工程，2023,38(9):152-157.LI Ligang,SUN Liqun,DAI Yongshou,et al.Ultra-shortterm prediction method of photovoltaic power based on spatio-temporal graph convolutional neural network[J].Journal of Engineering for Thermal Energy and Power,2023,38(9):152-157.
• [20]苏向敬，聂良钊，李超杰，等.基于MSTAGNN模型的可解释海上风电场多风机出力预测[J].电力系统自动化，2023,47(9):88-98.SU Xiangjing,NIE Liangzhao,LI Chaojie,et al.Interpretable power output prediction of multiple wind turbines for offshore wind farm based on multiple spatio-temporal attention graph neural network model[J]. Automation of Electric Power Systems,2023,47(9):88-98.
• [21] SIMEUNOVI?J,SCHUBNEL B,ALET P J,et al.Spatio-temporal graph neural networks for multi-site PV power forecasting[J]. IEEE Transactions on Sustainable Energy,2022,13(2):1210-1220.
• [22]高博，茆超，张冲标，等.基于时空图网络的分布式光伏发电出力预测[J].电力系统及其自动化学报，2023,35(3):125-133.GAO Bo,MAO Chao,ZHANG Chongbiao,et al.Output prediction of distributed photovoltaic power generation based on spatialtemporal graph neural network[J]. Proceedings of the CSU-EPSA,2023,35(3):125-133.
• [23]王玉庆，徐飞，刘志坚，等.基于动态关联表征与图网络建模的分布式光伏超短期功率预测[J].电力系统自动化，2023,47(20):72-82.WANG Yuqing,XU Fei,LIU Zhijian,et al. Ultra-shortterm power forecasting of distributed photovoltaic based on dynamic correlation characterization and graph network modeling[J]. Automation of Electric Power Systems,2023,47(20):72-82.
• [24]郑毅，王承民，刘保良，等.基于多层级时空图神经网络的风电机组在线异常检测[J].电力系统自动化，2024,48(5):107-119.ZHENG Yi,WANG Chengmin,LIU Baoliang,et al.Online anomaly detection of wind turbines based on hierarchical spatio-temporal graph neural network[J].Automation of Electric Power Systems,2024,48(5):107-119.
• [25]张延宇，张智铭，刘春阳，等.基于动态自适应图神经网络的电动汽车充电负荷预测[J].电力系统自动化，2024,48(7):86-93.ZHANG Yanyu,ZHANG Zhiming,LIU Chunyang,et al.Electric vehicle charging load prediction based on dynamic adaptive graph neural network[J].Automation of Electric Power Systems,2024,48(7):86-93.
• [26] OLUWASANMI A,AFTAB M U,QIN Z G,et al.Multihead spatiotemporal attention graph convolutional network for traffic prediction[J].Sensors,2023,23(8):3836.