王本涛,白杨,邢红涛,徐岩

• 1:国网呼伦贝尔供电公司
• 2:河北冀研能源科学技术研究院有限公司
• 3:华北电力大学电力工程系

摘要(Abstract)：

随着光伏并网容量的不断增加，准确的光伏功率预测对电网安全稳定运行意义重大。本文提出一种基于季节性分解与MMoE多任务学习的区域多光伏电站超短期功率联合预测方法。首先，通过季节性分解获得光伏功率的周期分量、剩余分量与趋势分量。其次，提出MMoE-LSTM-Attention网络来挖掘同一区域内不同光伏电站剩余分量与趋势分量之间的相关性，进行剩余分量与趋势分量的预测。最后，将分量进行汇总，得到光伏电站超短期功率预测结果。相较于传统基于硬共享机制的多任务学习模型，MMoE模型能够自动调整任务目标和任务间关系的参数权重。注意力机制能够进一步优化子任务的特征提取能力。在DKASC数据集上进行了算例实测，分别验证了季节性分解、MMoE多任务学习模型及注意力机制在区域多光伏电站功率预测问题上的有效性。

关键词(KeyWords)： 区域光伏功率预测;MMoE多任务学习;注意力机制;季节性分解

Abstract:

Keywords:

基金项目(Foundation): 国家自然科学基金资助项目（51307059）;; 2021年度石家庄重点研发计划资助项目（20300-装备制造创新专项）

作者(Author): 王本涛,白杨,邢红涛,徐岩

DOI: 10.19635/j.cnki.csu-epsa.000930

参考文献(References)：

• [1]龚莺飞，鲁宗相，乔颖，等(Gong Yingfei,Lu Zongxiang,Qiao Ying,et al).光伏功率预测技术(An overview of photovoltaic energy system output forecasting technology)[J].电力系统自动化(Automation of Electric Power Systems),2016,40(4):140-151.
• [2]赵亮，黎嘉明，艾小猛，等(Zhao Liang,Li Jiaming,Ai Xiaomeng,et al).光伏出力随机性分量的提取和统计特性分析(Analysis on random component extraction and statistical characteristics of photovoltaic power)[J].电力系统自动化(Automation of Electric Power Systems),2017,41(1):48-56.
• [3]刘杰，陈雪梅，陆超，等(Liu Jie,Chen Xuemei,Lu Chao,et al).基于数据统计特性考虑误差修正的两阶段光伏功率预测(Two-stage photovoltaic power forecasting and error correction method based on statistical characteristics of data)[J].电网技术(Power System Technology),2020,44(8):2891-2897.
• [4]Wang Zhengxin,Wang Zhiwei,Li Qin.Forecasting the industrial solar energy consumption using a novel seasonal GM(1,1)model with dynamic seasonal adjustment factors[J].Energy,2020,200:117460.
• [5]王佶宣，邓斌，王江(Wang Jixuan,Deng Bin,Wang Jiang).基于经验模态分解与RBF神经网络的短期风功率预测(Short-term wind power prediction based on empirical mode decomposition and RBF neural network)[J].电力系统及其自动化学报(Proceedings of the CSU-EP-SA),2020,32(11):109-115.
• [6]余向阳，赵怡茗，杨宁宁，等(Yu Xiangyang,Zhao Yiming,Yang Ningning,et al).基于VMD-SE-LSSVM和迭代误差修正的光伏发电功率预测(Photovoltaic power generation forecasting based on VMD-SE-LSSVM and iterative error correction)[J].太阳能学报(Acta Energiae Solaris Sinica),2020,41(2):310-318.
• [7]张雲钦，程起泽，蒋文杰，等(Zhang Yunqin,Cheng Qize,Jiang Wenjie,et al).基于EMD-PCA-LSTM的光伏功率预测模型(Photovoltaic power prediction model based on EMD-PCA-LSTM)[J].太阳能学报(Acta Energiae Solaris Sinica),2021,42(9):62-69.
• [8]杨国清，张凯，王德意，等(Yang Guoqing,Zhang Kai,Wang Deyi,et al).基于包络线聚类的多模融合超短期光伏功率预测算法(Multi-mode fusion ultra-short-term photovoltaic power prediction algorithm based on envelope clustering)[J].电力自动化设备(Electric Power Automation Equipment),2021,41(2):39-46.
• [9]叶林，裴铭，路朋，等(Ye Lin,Pei Ming,Lu Peng,et al).基于天气分型的短期光伏功率组合预测方法(Combination forecasting method of short-term photovoltaic power based on weather classification)[J].电力系统自动化(Automation of Electric Power Systems),2021,45(1):44-54.
• [10]吉锌格，李慧，刘思嘉，等(Ji Xinge,Li Hui,Liu Sijia,et al).基于MIE-LSTM的短期光伏功率预测(Short-term photovoltaic power forecasting based on MIE-LSTM)[J].电力系统保护与控制(Power System Protection and Control),2020,48(7):50-57.
• [11]王育飞，付玉超，薛花(Wang Yufei,Fu Yuchao,Xue Hua).计及太阳辐射和混沌特征提取的光伏发电功率DMCS-WNN预测法(DMCS-WNN prediction method of photovoltaic power generation by considering solar radiation and chaotic feature extraction)[J].中国电机工程学报(Proceedings of the CSEE),2019,39(增)：63-71.
• [12]Shaker H,Manfre D,Zareipour H.Forecasting the aggregated output of a large fleet of small behind-the-meter solar photovoltaic sites[J].Renewable Energy,2020,147:1861-1869.
• [13]Qu Zhijian,Xu Juan,Wang Zixiao,et al.Prediction of electricity generation from a combined cycle power plant based on a stacking ensemble and its hyperparameter optimization with a grid-search method[J].Energy,2021,227:120309.
• [14]Vaswani A,Shazeer N,Parmar N,et al.Attention is all you need[C]//Conference on Advances in Neural Information Processing Systems.Long Beach,USA,2017:5999-6009.
• [15]胡兵，詹仲强，陈洁，等(Hu Bing,Zhan Zhongqiang,Chen Jie,et al).基于PCA-GA-Elman的短期光伏出力预测研究(Prediction research on short-term photovoltaic output based on PCA-GA-Elman)[J].太阳能学报(Acta Energiae Solaris Sinica),2020,41(6):256-263.
• [16]孟安波，许炫淙，陈嘉铭，等(Meng Anbo,Xu Xuancong,Chen Jiaming,et al).基于强化学习和组合式深度学习模型的超短期光伏功率预测(Prediction of ultra-shortterm photovoltaic power based on reinforcement learning and Combined deep learning model)[J].电网技术(Power System Technology),2021,45(12):4721-4728.
• [17]周楠，徐潇源，严正，等(Zhou Nan,Xu Xiaoyuan,Yan Zheng,et al).基于宽度学习系统的光伏发电功率超短期预测(Ultra-short-term forecasting of photovoltaic power generation based on broad learning system)[J].电力系统自动化(Automation of Electric Power Systems),2021,45(1):55-64.
• [18]Ma Jiaqi,Zhao Zhe,Yi Xinyang,et al.Modeling task relationships in multi-task learning with multi-gate mixture-ofexperts[C]//ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.London,UK,2018:1930-1939.
• [19]DKA Solar Centre.DKASC,Alice Springs[EB/OL].http：//dkasolarcentre.com.au/locations/alice-springs,2020.