陈哲,周金辉,靳东辉,陈积光,马恒瑞,张嘉鑫,朱苏洵
CHEN Zhe,ZHOU Jinhui,JIN Donghui,CHEN Jiguang,MA Hengrui,ZHANG Jiaxin,ZHU Suxun

• 1:国网浙江省电力有限公司
• 2:国网浙江省电力有限公司瑞安市供电公司
• 3:交直流智能配电网湖北省工程中心
• 4:武汉大学电气与自动化学院

摘要(Abstract)：

提出一种基于TimeVAE（时间变分自动编码器）和TL（迁移学习）的集成学习方法，用于解决IES（综合能源系统）中新建系统因历史数据稀缺导致负荷预测精度下降的问题。通过Time VAE的变分自编码器生成负荷数据，增强目标域数据集的多样性；利用具有丰富历史数据的源域知识，通过冻结训练策略优化目标域负荷预测模型。基于亚利桑那州立大学不同校区IES的数据，验证了所提方法的有效性。实验结果表明，该方法在小样本情况下显著提升了负荷预测的精度，为实现IES负荷预测的高效性和可靠性提供了重要参考。
In this paper, an integrated learning approach based on time variational autoencoder(TimeVAE) and transfer learning(TL) is proposed to address the accuracy degradation in integrated energy systems(IESs) load forecasting caused by insufficient historical data in newly built systems. The TimeVAE-based variational autoencoder generates load data to enhance the diversity of the target-domain dataset, while the source-domain knowledge with abundant historical data is leveraged to optimize the target-domain load forecasting model through a frozen training strategy. Based on IES data from multiple campuses of Arizona State University, the effectiveness of the proposed method is validated. Experimental results demonstrate that the method significantly improves load forecasting accuracy under few-shot conditions, providing critical references for achieving efficient and reliable IES load forecasting.

关键词(KeyWords)： TimeVAE;迁移学习;小样本负荷预测;综合能源系统;数据增强;冻结训练策略;多能负荷预测
TimeVAE;transfer learning;few-shot load forecasting;IES;data augmentation;frozen training strategy;multi-energy load forecasting

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金(61933005);; 国网浙江省电力有限公司科技项目(2006CB200303)

作者(Author): 陈哲,周金辉,靳东辉,陈积光,马恒瑞,张嘉鑫,朱苏洵
CHEN Zhe,ZHOU Jinhui,JIN Donghui,CHEN Jiguang,MA Hengrui,ZHANG Jiaxin,ZHU Suxun

DOI: 10.19585/j.zjdl.202509006

参考文献(References)：

• [1]雷杨，赵纪峰，丁石川，等.考虑双重不确定性的区域综合能源系统多阶段滚动随机规划[J].电力系统自动化，2023,47(20):53-63.LEI Yang,ZHAO Jifeng,DING Shichuan,et al. Multistage rolling stochastic planning of regional integrated energy system considering dual-uncertainty[J]. Automation of Electric Power Systems,2023,47(20):53-63.
• [2] WU J,YAN J Y,JIA H J,et al.Integrated energy systems[J].Applied Energy,2016,167:155-157.
• [3]王婷，张晶，高冲，等.绿证-碳市场互认抵消下综合能源系统优化调度[J].智慧电力，2025,53(2):104-110.WANG Ting,ZHANG Jing,GAO Chong,et al.Optimal scheduling of integrated energy system under mutual recognition and offset of green certificate and carbon market[J].Smart Power,2025,53(2):104-110.
• [4]鲍国栋，宋柯，钱定冬，等.极性反转条件下XLPE空间电荷与电场分布特性研究[J].高压电器，2025,61(8):146-154.BAO Guodong,SONG Ke,QIAN Dingdong,et al.Study on space charge and electric field distribution characteristics of XLPE under polarity reversal voltage[J].High Voltage Apparatus,2025,61(8):146-154.
• [5]束娜，江山，刘春伶，等.计及灵活性资源多时间尺度协调互济的电-气-热综合能源系统优化调度[J].电力建设，2024,45(12):3-15.SHU Na,JIANG Shan,LIU Chunling,et al. Optimal scheduling of electricity-gas-heat integrated energy system with flexible resources in multiple time scales[J].Electric Power Construction,2024,45(12):3-15.
• [6]李欣，陈英彰，李涵文，等.考虑碳交易的电-热综合能源系统两阶段鲁棒优化低碳经济调度[J].电力建设，2024,45(6):58-69.LI Xin,CHEN Yingzhang,LI Hanwen,et al. Two-stage robust optimization of low-carbon economic dispatch for electricity-thermal integrated energy system considering carbon trade[J]. Electric Power Construction,2024,45(6):58-69.
• [7]潘廷哲，王宗义，杨晨，等.面向电力需求响应的工业园区综合能源系统优化调控研究[J].电力需求侧管理，2024,26(6):62-67.PAN Tingzhe,WANG Zongyi,YANG Chen,et al. Research on optimal dispatching of industrial parks integrated energy system for demand-side response[J]. Power Demand Side Management,2024,26(6):62-67.
• [8]朱海南，李丰硕，孙华忠，等.基于改进AlexNet-GRU深度学习网络的配电网短期负荷预测方法[J].电力电容器与无功补偿，2023,44(4):48-54.ZHU Hainan,LI Fengshuo,SUN Huazhong,et al.Shortterm load prediction method of distribution network based on improved AlexNet-GRU deep learning network[J].Power Capacitor&Reactive Power Compensation,2023,44(4):48-54.
• [9] ZHU J,DONG H,LI S,et al.Review of data-driven load forecasting for integrated energy system[J]. Proceedings of the CSEE,2021,41(23):7905-7924.
• [10]欧锋，罗醒华，龙经纬，等.基于图转换和迁移学习的低压配电网户变关系和相位识别方法[J].浙江电力，2024,43(3):95-103.OU Feng,LUO Xinghua,LONG Jingwei,et al. A household-transformer relationships and phase identification method of low-voltage distribution networks based on graph transformation and transfer learning[J]. Zhejiang Electric Power,2024,43(3):95-103.
• [11] YUAN Y,CHEN Z H,WANG Z,et al. Attention mechanism-based transfer learning model for day-ahead energy demand forecasting of shopping mall buildings[J].Energy,2023,270:126878.
• [12] PENG C,TAO Y F,CHEN Z P,et al.Multi-source transfer learning guided ensemble LSTM for building multiload forecasting[J]. Expert Systems with Applications,2022,202:117194.
• [13] CAI L,GU J,JIN Z J.Two-layer transfer-learning-based architecture for short-term load forecasting[J]. IEEE Transactions on Industrial Informatics,2020,16(3):1722-1732.
• [14] XU X Z,MENG Z R.A hybrid transfer learning model for short-term electric load forecasting[J].Electrical Engineering,2020,102(3):1371-1381.
• [15] LU Y K,TIAN Z,ZHOU R Y,et al.A general transfer learning-based framework for thermal load prediction in regional energy system[J].Energy,2021,217:119322.
• [16] WANG C D,YUAN J J,HUANG K,et al.Research on thermal load prediction of district heating station based on transfer learning[J].Energy,2022,239:122309.
• [17] LI C,LI G J,WANG K Y,et al.A multi-energy load forecasting method based on parallel architecture CNN-GRU and transfer learning for data deficient integrated energy systems[J].Energy,2022,259:124967.
• [18] WEI N,YIN C,YIN L H,et al.Short-term load forecasting based on WM algorithm and transfer learning model[J].Applied Energy,2024,353:122087.
• [19]韩睿，戴哲仁，蒋鹏，等.基于改进YOLOv8的电力场景通用缺陷检测模型[J].浙江电力，2024,43(4):113-120.HAN Rui,DAI Zheren,JIANG Peng,et al.A general defect detection model for power scenarios using the improved YOLOv8[J]. Zhejiang Electric Power,2024,43(4):113-120.
• [20] SU J N,CHEN F C,QIU Z J,et al.Multi-load short-term prediction of an integrated energy system based on GANLSTM[J]. Journal of Physics:Conference Series,2023,2564(1):012062.
• [21]曾爽，丁屹峰，李香龙，等.基于条件时序生成对抗网络的楼宇空调负荷场景生成方法[J].电网技术，2022,46(7):2448-2456.ZENG Shuang,DING Yifeng,LI Xianglong,et al. Scenario generation of air conditioning loads in buildings based on conditional TimeGAN[J].Power System Technology,2022,46(7):2448-2456.
• [22]何宇浩，宋云海，何森，等.面向电力缺陷场景的小样本图像生成方法[J].浙江电力，2024,43(1):126-132.HE Yuhao,SONG Yunhai,HE Sen,et al.A few-shot image generation method for power defect scenarios[J].Zhejiang Electric Power,2024,43(1):126-132.
• [23]梁俊杰，韦舰晶，蒋正锋.生成对抗网络GAN综述[J].计算机科学与探索，2020,14(1):1-17.LIANG Junjie,WEI Jianjing,JIANG Zhengfeng.Generative adversarial networks GAN overview[J]. Journal of Frontiers of Computer Science and Technology,2020,14(1):1-17.
• [24] LAN J,ZHOU Y Z,GUO Q L,et al.Data augmentation for data-driven methods in power system operation:a novel framework using improved GAN and transfer learning[J]. IEEE Transactions on Power Systems,2024,39(5):6399-6411.
• [25] LU Y T,WANG G C,HUANG S Q.A short-term load forecasting model based on mixup and transfer learning[J].Electric Power Systems Research,2022,207:107837.
• [26] ZHOU D J,MA S X,HAO J R,et al.An electricity load forecasting model for integrated energy system based on BiGAN and transfer learning[J].Energy Reports,2020,6:3446-3461.
• [27] SU J J,YU X J,WANG X R,et al. Enhanced transfer learning with data augmentation[J].Engineering Applications of Artificial Intelligence,2024,129:107602.
• [28] LI K J,WEI B R,TANG Q Q,et al.A data-efficient building electricity load forecasting method based on maximum mean discrepancy and improved TrAdaBoost algorithm[J].Energies,2022,15(23):8780.
• [29] DESAI A,FREEMAN C,WANG Z H,et al.TimeVAE:a variational auto-encoder for multivariate time series generation[EB/OL]. 2021:2111.08095. https：//arxiv. org/abs/2111.08095v3.
• [30] ZHU S X,MA H R,CHEN L J,et al. Short-term load forecasting of an integrated energy system based on STLCPLE with multitask learning[J].Protection and Control of Modern Power Systems,2024,9(6):71-92.
• [31] LIASHCHYNSKYI P,LIASHCHYNSKYI P. Grid search,random search,genetic algorithm:a big comparison for NAS[EB/OL]. 2019:1912.06059. https：//arxiv.org/abs/1912.06059v1.
• [32]李云松，张智晟.考虑综合需求响应的Trans-GNN综合能源系统多元负荷短期预测[J/OL].电工技术学报，2023:1-11.(2023-12-28). https：//kns. cnki. net/KCMS/detail/detail. aspx? filename=DGJS20231226003&dbname=CJFD&dbcode=CJFQ.LI Yunsong,ZHANG Zhisheng.Short-term forecasting of multi-load of Trans-GNN integrated energy system considering comprehensive demand response[J/OL]. China Industrial Economics,2023:1-11.(2023-12-28). https：//kns. cnki. net/KCMS/detail/detail. aspx? filename=DGJS20231226003&dbname=CJFD&dbcode=CJFQ.
• [33]鲁斌，霍泽健，俞敏.基于LSTNet-Skip的综合能源系统多元负荷超短期预测[J].中国电机工程学报，2023,43(6):2273-2283.LU Bin,HUO Zejian,YU Min.Multi load ultra short-term forecasting of integrated energy system based on LSTNetskip[J]. Proceedings of the CSEE,2023,43(6):2273-2283.
• [34] NIU D X,YU M,SUN L J,et al.Short-term multi-energy load forecasting for integrated energy systems based on CNN-BiGRU optimized by attention mechanism[J]. Applied Energy,2022,313:118801.
• [35]吴晨，姚菁，薛贵元，等.基于MMoE多任务学习和长短时记忆网络的综合能源系统负荷预测[J].电力自动化设备，2022,42(7):33-39.WU Chen,YAO Jing,XUE Guiyuan,et al.Load forecasting of integrated energy system based on MMoE multitask learning and LSTM[J]. Electric Power Automation Equipment,2022,42(7):33-39.