考虑可再生能源绿证交易的LSTNet 虚拟电厂负荷预测

NPS

Carbon Neutralization and New Power Systems

2995-44362995-4479

Art and Design

10.61369/NPS.2025020002

Article

考虑可再生能源绿证交易的LSTNet 虚拟电厂负荷预测https://artdesignp.com/journal/NPS/3/2/10.61369/NPS.2025020002张宜然,姜晓霞,白宁,高康伟,李芳菲,黄一峻,张艳灵

2025

2025-06-20

【目的】旨在解决可再生能源绿证交易背景下虚拟电厂中长期负荷预测的精度问题，针对数据缺失、算法运行时间长等挑战，重点探究绿证交易对负荷变化的影响机理，并优化预测模型以提升精准度。【方法】首先通过皮尔逊相关性分析验证负荷与非同步发电瞬时渗透率（SNSP）的中等相关性（相关系数0.410），证明绿证交易对负荷存在影响；利用密度聚类算法（DBSCAN）提取负荷季节性特征（分为春夏/ 秋冬两类），缩减训练数据规模。在此基础上，提出DBSCAN- LSTNet 混合预测模型：采用一维CNN 提取短期时间序列特征，结合GRU 和Skip-GRU 捕获长周期依赖关系，并通过自回归模块（AR）解决非线性特征导致的尺度敏感性问题。以SMAPE 为评价指标，使用北爱尔兰2018 年–2020 年负荷数据进行训练和验算，并引入SNSP 表征绿证交易强度。【结果】实验表明：（1）考虑绿证交易因素（SNSP）的DBSCAN-LSTNet 模型误差降至2.56%（未考虑时为6.03%），显著优于传统的LSTM（3.91%）和SVM（23.45%）；（2）绿证因素可使预测误差平均降低4%；（3）DBSCAN 有效缩减数据规模，模型训练效率提升，且对离群点具有鲁棒性；（4）LSTNet 融合线性和非线性预测，比单一LSTM 具有更高精度与鲁棒性。【结论】虚拟电厂负荷预测需纳入绿证交易等市场因素。所提DBSCAN- LSTNet 模型通过特征降维和混合神经网络结构，实现了高精度中长期负荷预测（SMAPE≤2.56%），为电力市场决策提供可靠依据。虚拟电厂,中长期负荷预测,绿证交易,神经网络, DBSCAN-LSTNet 混合预测模型

[1]XU X Y,YAN Z,SHAHIDEHPOUR M,et al.Data-Driven Risk-Averse Two-Stage Optimal Stochastic Scheduling of Energy and Reserve with Correlated Wind Power[J]. IEEE Transactions on Sustainable Energy, 2020,11(1): 436-447. [2] WANG C,LIU F,WANG J H,et al.Robust Risk-Constr- ained Unit Commitment with Large-Scale Wind Gene- ration:an Adjustable Uncertainty Set Approach[J].IEEE Transactions on Power Systems,2017,32 (1):723-733. [3] WEI W,LIU F,MEI S W.Distributionally Robust Co-Opt- imization of Energy and Reserve Dispatch[J].IEEE Tr- ansactions on Sustainable Energy,2016,7(1):289-300. [4] PARAG Y,SOVACOOL B K.Electricity Market Design for the Prosumer Era[J].Nature Energy,2016,1:16032. [5] 谈金晶,李扬.多能源协同的交易模式研究综述[J].中国电机工程学报,2019,39(22):6483-6497.TAN Jinjing,LI Yang.Review on Transaction Mode in Multi-Energy Collaborative Market[J].Proceedings of the CSEE,2019,39(22):6483-6497. [6] TASCIKARAOGLU A,ERDINC O,UZUNOGLU M,et al.An Adaptive Load Dispatching and Forecasting Stra- tegy for a Virtual Power Plant Including Renewable Energy Conversion Units[J].Applied Energy,2014,119: 445-453. [7] CHEN Y,LI T X,ZHAO C H,et al.Decentralized Provi- sion of Renewable Predictions within a Virtual Power Plant[J].IEEE Transactions on Power Systems,2021,36 (3):2652-2662. [8] MORENO G,MARTIN P,SANTOS C,et al.A Day-Ahead Irradiance Forecasting Strategy for the Integration of Photovoltaic Systems in Virtual Power Plants[J].IEEE Access,2020,8:204226-204240. [9] HERNANDEZ L,BALADRON C,AGUIAR J M,et al.A Multi-Agent System Architecture for Smart Grid Man- agement and Forecasting of Energy Demand in Virtual Power Plants[J].IEEE Communications Magazine,2013, 51(1):106-113. [10] 李鹏,何帅,韩鹏飞,等.基于长短期记忆的实时电价条件下智能电网短期负荷预测[J].电网技术,2018,42 (12):4045-4052.LI Peng,HE Shuai,HAN Pengfei,et al.Short-Term Load Forecasting of Smart Grid Based on Long-Short-Term Memory Recurrent Neural Networks in Condition of Real-Time Electricity Price[J].Power System Technol- ogy,2018,42(12):4045-4052. [11] BENGIO Y,COURVILLE A,VINCENT P.Represen- tation Learning:a Review and New Perspectives[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2013,35(8):1798-1828. [12] KONG W C,DONG Z Y,JIA Y W,et al.Short-Term Residential Load Forecasting Based on LSTM Rec- urrent Neural Network[J].IEEE Transactions on Smart Grid,2019,10(1):841-851. [13] 郭亦宗,冯斌,岳铂雄,等.负荷聚合商模式下考虑需求响应的超短期负荷预测[J].电力系统自动化,2021, 45(1):79-87.GUO Yizong,FENG Bin,YUE Boxiong,et al.Ultra- Short-Term Load Forecasting Considering Demand Response in Load Aggregator Mode[J].Automation of Electric Power Systems,2021,45(1):79-87. [14] CHEN Y T,ZHANG D X.Theory-Guided Deep-Lear- ning for Electrical Load Forecasting (TgDLF) via Ensemble Long Short-Term Memory[J].Advances in Applied Energy,2021,1:100004. [15] CHEN K J,CHEN K L,WANG Q,et al.Short-Term Load Forecasting with Deep Residual Networks[J].IEEE Tra- nsactions on Smart Grid,2019,10(4):3943-3952. [16] LAI G K,CHANG W C,YANG Y M,et al.Modeling Long and Short-Term Temporal Patterns with Deep Neural Networks[C]//The 41st International ACMSI- GIR Conference on Research & Development in Infor- mation Retrieval.Ann Arbor MI USA.ACM,2018: 95- 104. [17] 樊宇琦,丁涛,孙瑜歌,等.国内外促进可再生能源消纳的电力现货市场发展综述与思考[J].中国电机工程学报,2021,41(5):1729-1752.FAN Yuqi,DING Tao,SUN Yuge,et al.Review and Cog- itation for Worldwide Spot Market Development to Promote Renewable Energy Accommodation[J].Pro- ceedings of the CSEE,2021,41(5):1729-1752. [18] 单茂华,汤洪海,耿明志,等.绿色电力市场本质动因及设计思考[J].电力系统自动化,2020,44(16):12-20.SHAN Maohua,TANG Honghai,GENG Mingzhi,et al. Essential Cause and Design Thinking of Green Electr- icity Market[J].Automation of Electric Power Systems, 2020,44(16):12-20. [19] GUO H Y,CHEN Q X,XIA Q,et al.Modeling Strategic Behaviors of Renewable Energy with Joint Consider- ation on Energy and Tradable Green Certificate Mark- ets[J].IEEE Transactions on Power Systems,2020,35 (3):1898-1910. [20] BUTLER L,NEUHOFF K.Comparison of Feed-in Tariff,Quota and Auction Mechanisms to Support Wind Power Development[J].Renewable Energy,2008,33(8): 1854-1867. [21] AUNE F R,DALEN H M,HAGEM C.Implementing the EU Renewable Target through Green Certificate Mark- ets[J].Energy Economics,2012,34(4):992-1000. [22] AUNE F R,DALEN H M,HAGEM C.Implementing the EU Renewable Target through Green Certificate Mark- ets[J].Energy Economics,2012,34(4):992-1000. [23] 张悦,谢敏,程培军,等.可再生能源绿证价格季节性测算方法研究[J].南方能源建设,2020,7(3):46-54.ZHANG Yue,XIE Min,CHENG Peijun,et al.Research on Seasonal Calculation Method of Renewable Energy Certificate Price[J].Southern Energy Construction,2020, 7(3):46-54. [24] ZHU Q D,TANG X M,LIU Z L.Revised DBSCAN Clustering Algorithm Based on Dual Grid[C]//2020 Chinese Control and Decision Conference (CCDC). August 22-24,2020,Hefei,China.IEEE,2020:3461-3466. [25] KOUKARAS P,BEZAS N,GKAIDATZIS P,et al.Intro- ducing a Novel Approach in One-Step Ahead Energy Load Forecasting[J].Sustainable Computing:Informa- tics and Systems,2021,32:100616. [26] MOTEPE S,HASAN A N,TWALA B,et al.Power Distr- ibution Networks Load Forecasting Using Deep Belief Networks:The South African Case[C]//2019 IEEE Jor- dan International Joint Conference on Electrical Engin- eering and Information Technology (JEEIT).April 9-11, 2019,Amman,Jordan.IEEE,2019:507-512.