陈以恒,杨秀,孙改平,杨帆
CHEN Yiheng,YANG Xiu,SUN Gaiping,YANG Fan

• 1:上海电力大学电气工程学部
• 2:国家电网上海电力科学研究院

摘要(Abstract)：

当前配电变压器（以下简称“配变”）的台区电压预测研究多以时间序列建模的单点预测为主，对配变间复杂的时空相关性建模不充分，大规模配变集群场景下的扩展性和计算效率存在不足。为此，提出一种基于长短时间尺度双层聚类与时空图注意力网络的电压预测方法。首先，在短时间尺度上刻画典型日的电压波动特征，构建运行特性相似的配变子集群；其次，在长时间尺度上提取年运行水平特征，形成双层聚类结构；在此基础上，构建融合电压、有功功率及地理位置三类异构图结构的多通道图注意力网络，实现对集群电压的精准预测。最后，基于上海市某区域配变的真实运行数据进行算例仿真，结果验证了该方法在提升配变电压预测精度方面的有效性。
Existing research on voltage prediction for distribution transformers(hereinafter referred to as “DTs”) primarily relies on single-point time-series modeling, which insufficiently captures complex spatiotemporal correlations among DTs. This leads to limited scalability and computational efficiency in large-scale DT cluster scenarios. To address these issues, a voltage prediction method based on dual-layer clustering across long-and short-term time scales and a spatiotemporal graph attention network is proposed. First, on a short-term time scale, voltage fluctuation characteristics of typical daily profiles are extracted to construct sub-clusters of DTs with similar operating behaviors. Second, on a long-term time scale, annual operational features are extracted to form a dual-layer clustering structure. On this basis, a multi-channel graph attention network is developed by integrating heterogeneous graph structures of voltage, active power, and geographical location, enabling accurate prediction of cluster-level voltage. Finally, case studies using real operational data from a distribution transformer in Shanghai validate the effectiveness of the proposed method in improving voltage prediction accuracy for distribution transformers.

关键词(KeyWords)： 配变电压预测;多时间尺度特征;双层聚类;时空相关性;图神经网络;多通道注意力机制
distribution transformer voltage prediction;multi-time-scale features;dual-layer clustering;spatiotemporal correlation;graph neural network;multi-channel attention mechanism

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金（52177098）;; 上海电力人工智能工程技术研究中心项目（19DZ2252800）

作者(Author): 陈以恒,杨秀,孙改平,杨帆
CHEN Yiheng,YANG Xiu,SUN Gaiping,YANG Fan

DOI: 10.19585/j.zjdl.202609008

参考文献(References)：

• [1]倪淏，戴聪，褚红雷，等.分布式能源场景下电能质量问题及其调控策略综述[J].电工技术，2024(4):169-172.Ni Hao,Dai Cong,Chu Honglei,et al.Overview of power quality issues and control strategies in the environment of distributed sources[J]. Electric Engineering,2024(4):169-172.
• [2]师伟，马雨姣，姬雨婷.新能源发电并网对电网质量的影响和应对策略[J].能源与节能，2025(11):46-49.Shi Wei,Ma Yujiao,Ji Yuting. Impact of new energy power generation grid connection on power grid quality and corresponding countermeasures[J]. Energy and Conservation,2025(11):46-49.
• [3]Yuan W Q,Yuan X,Xu L W,et al.Harmonic loss analysis of low-voltage distribution network integrated with distributed photovoltaic[J].Sustainability,2023,15(5):4334.
• [4]柳明，乔磊.分布式电源接入对配网供电质量的影响及优化策略[J].产品可靠性报告，2025(11):104-106.Liu Ming,Qiao Lei.Influence of distributed power supply access on power supply quality of distribution network and its optimization strategy[J]. Automotive Quality,2025(11):104-106.
• [5]崔柳，李成豪，谢赟，等.基于光伏发电预测的分布式协同电压控制方法[J].可再生能源，2025,43(8):1106-1113.Cui Liu,Li Chenghao,Xie Yun,et al.Distributed cooperative voltage control method based on PV power generation prediction[J].Renewable Energy Resources,2025,43(8):1106-1113.
• [6]赵长伟，骈睿珺，杜天硕，等.基于LSTM的重要用户电能质量趋势预测分析模型[J].电力系统及其自动化学报，2022,34(7):26-33.Zhao Changwei,Pian Ruijun,Du Tianshuo,et al. Trend prediction and analysis model for power quality of important users based on LSTM[J].Proceedings of the CSUEPSA,2022,34(7):26-33.
• [7]陈茜，白晶，陈小月，等.分布式光伏并网对配电网的影响综述[J].科学技术与工程，2024,24(27):11491-11504.Chen Qian,Bai Jing,Chen Xiaoyue,et al. Review of the impact of distributed PV grid integration on the power grid[J].Science Technology and Engineering,2024,24(27):11491-11504.
• [8]Gulraiz A,Zaidi S S H,Gulraiz H,et al.Leveraging dense layer hybrid graph neural networks for managing overvoltage in PV-dominated distribution systems[J]. Results in Engineering,2025,27:106169.
• [9]卢锦玲，任义杰，李承远，等.基于动态时空预测的多时间尺度配电网电压优化[J].电力科学与工程，2025,41(7):1-9.Lu Jinling,Ren Yijie,Li Chengyuan,et al. Multi-timescale voltage optimization of distribution networks based on dynamic spatio-temporal prediction[J].Electric Power Science and Engineering,2025,41(7):1-9.
• [10]Zhang X,Wu Z,Sun Q R,et al.Application and progress of artificial intelligence technology in the field of distribution network voltage Control:a review[J].Renewable and Sustainable Energy Reviews,2024,192:114282.
• [11]尹忠东，牟锴，金涌涛，等.基于BP神经网络的低压配电台区电压估算[J].华北电力大学学报（自然科学版），2017,44(5):27-33.Yin Zhongdong,Mu Kai,Jin Yongtao,et al.Voltage estimation of low-voltage distribution network based on BP neural network[J].Journal of North China Electric Power University(Natural Science Edition),2017,44(5):27-33.
• [12]韩雨，郭成，方正云，等.基于AP-WOA-GRU的分布式光伏集群电压越限动态预测[J].电网与清洁能源，2024,40(4):118-126.Han Yu,Guo Cheng,Fang Zhengyun,et al.The dynamic prediction of voltage overrun of the distributed photovoltaic cluster based on AP-WOA-GRU[J]. Advances of Power System&Hydroelectric Engineering,2024,40(4):118-126.
• [13]Wang Y,Von Krannichfeldt L,Zufferey T,et al. Shortterm nodal voltage forecasting for power distribution grids:an ensemble learning approach[J].Applied Energy,2021,304:117880.
• [14]Yi C Y,Zhou Y J,Huang W X.A low voltage prediction based on LSTM-BP combined model for distribution station areas[J].Journal of Physics:Conference Series,2023,2584(1):012120.
• [15]朱三立，李娜娜，徐俊俊，等.计及随机扰动信号分析的有源配电网模型预测电压控制[J/OL].电气工程学报，2025:1-11[2025-12-17]. https：//link. cnki. net/urlid/10.1289.tm.20251014.1515.002.Zhu Sanli,Li Nana,Xu Junjun,et al. Predictive voltage control of active distribution network model considering random disturbance signal analysis[J/OL].Journal of Electrical Engineering,2025:1-11[2025-12-17].https：//link.cnki.net/urlid/10.1289.tm.20251014.1515.002.
• [16]周杨珺，张斌，黄伟翔，等.基于LSTM-BP组合模型的配电台区低电压预测[J].电力科学与技术学报，2023,38(5):177-186.Zhou Yangjun,Zhang Bin,Huang Weixiang,et al.A low voltage prediction based on LSTM-BP combined model for distribution station area[J].Journal of Electric Power Science and Technology,2023,38(5):177-186.
• [17]郭少东，赵晓莉，孙改平，等.基于关联特征筛选的双层聚类区域配变电压越限特征分析及预测[J].南方电网技术，2025,19(2):19-27.Guo Shaodong,Zhao Xiaoli,Sun Gaiping,et al. Analysis and prediction of over-limit characteristics for regional distribution transformer voltage using bilayer clustering by correlation feature screening[J]. Southern Power System Technology,2025,19(2):19-27.
• [18]魏金柱，田施兰，陈雯晨，等.基于GA-BP-LSTM模型的配电台区低电压预测[J].电工技术，2025(1):15-20.Wei Jinzhu,Tian Shilan,Chen Wenchen,et al.Low voltage prediction in distribution station area based on GA-BPLSTM model[J].Electric Engineering,2025(1):15-20.
• [19]Wang Z P,Lu Y Y,Wang B,et al.Terminal voltage prediction of distribution and substation based on spatialtemporal distribution characteristics of the station load[J].Journal of Nanoelectronics and Optoelectronics,2022,17(12):1554-1563.
• [20]Lyu Y L,Wang L,Long D H,et al.Multi-area short-term load forecasting based on spatiotemporal graph neural network[J]. Engineering Applications of Artificial Intelligence,2024,138:109398.
• [21]覃日升，张辰，马月，等.基于多神经网络混合模型的10kV配电台区低电压中长期预测[J].云南电力技术，2024,52(6):75-80.Qin Risheng,Zhang Chen,Ma Yue,et al.Low voltage medium to long-term prediction in 10 kV distribution station area based on multiple model[J].Yunnan Electric Power,2024,52(6):75-80.
• [22]盛冠群，吴航，范李平，等.基于电压特征谱聚类和EMA季节趋势分解网络的配变电压预测[J/OL].电力系统及其自动化学报，2025:1-11[2025-12-17].https：//doi.org/10.19635/j.cnki.csu-epsa.001739.Sheng Guanqun,Wu Hang,Fan Liping,et al.Distribution transformer voltage prediction based on voltage characteristic spectrum clustering and EMA seasonal trend decomposition network[J/OL].Proceedings of the CSU-EPSA,2025:1-11[2025-12-17].https：//doi.org/10.19635/j.cnki.csu-epsa.001739.