作者：卢忠山, 袁建华
作者单位：三峡大学电气与新能源学院，湖北 宜昌 443002
关键词：光伏发电;超短期;功率预测;模态分解
摘要：
为提高光伏发电系统功率超短期预测的准确性，提出一种基于EEMD-LSTM的光伏电站超短期预测模型。该模型选取某50 MW光伏电站2017年功率数据作为样本，根据天气因素分类指标将天气情况分为非突变天气和突变天气两大类，利用EEMD将分类天气的历史功率数据分解为IMF1~IMF5和剩余分量，计算各个分量与原始数据之间的相关性并将强相关的分量送入LSTM神经网络，叠加各子分量结果得到最终的光伏功率预测结果，同步搭建BP、SVM、KNN和LSTM模型与所提模型进行误差对比。结果表明：天气因素对光伏输出功率有较大影响；单一模型对功率波动较大的突变天气进行预测时会产生较大误差；功率数据经过EEMD分解，可充分提取细节特征，使得EEMD-LSTM耦合模型较LSTM模型在eRMSE、eMAPE、eTIC上分别提升21.23%、11.92%、25.67%。所提模型可有效提高光伏功率超短期预测的准确度，满足光伏发电系统超短期预测的要求。
Ultra-short term power prediction of photovoltaic power generation system based on EEMD-LSTM method
LU Zhongshan, YUAN Jianhua
College of Electrical and New Energy, China Three Gorges University, Yichang 443002, China
Abstract: An ultra-short-term power prediction model of photovoltaic power station based on EEMD-LSTM is proposed in order to improve the accuracy of ultra-short-term power prediction of photovoltaic power generation system. By taking the power data of a 50 MW photovoltaic power station in 2017 as a sample, the weather conditions are divided into two categories: non-abrupt weather and abrupt weather according to the classification index of weather factors. EEMD is used to decompose the historical power data of classified weather into IMF1-IMF5 and residual components. The correlation between each component and the original data is calculated, and the strongly correlated components are sent to LSTM neural network. The final photovoltaic power prediction results are obtained by superimposing the results of each sub-component. BP, SVM, KNN and LSTM models are built synchronously to compare the errors with the proposed models. The results show that weather factors have great influence on photovoltaic output power; When a single model predicts abrupt weather with large power fluctuation, it would produce large errors. After EEMD decomposition of power data, detail features can be fully extracted, which makes EEMD-LSTM coupling model improve by 21.23%, 11.92% and 25.67% on eRMSE, eMAPE and eTIC respectively compared with LSTM model. The proposed model effectively improves the accuracy of ultra-short-term prediction of PV power, and meets the requirements of ultra-short-term prediction of PV power generation system.
Keywords: photovoltaic power generation;super short term;power prediction;mode decomposition
2022, 48(12):125-132 收稿日期: 2021-07-13;收到修改稿日期: 2021-09-22
基金项目: 国家自然科学基金项目（61603212）
作者简介: 卢忠山(1992-),男,湖北宜昌市人,硕士研究生,专业方向为人工智能在电力系统的应用研究
参考文献
[1] 陈功, 许杨, 袁鑫, 等. 柔性光伏电池阵列电气参数建模研究[J]. 中国测试, 2020, 46(5): 134-142
[2] 吴坚, 郑照红, 薛家祥. 深度置信网络光伏发电短时功率预测研究[J]. 中国测试, 2018, 44(5): 6-11
[3] WANG H, LEI Z, ZHANG X, et al. A review of deep learning for renewable energy forecasting[J]. Energy Conversion and Management, 2019, 198: 111799
[4] 谢晓頔, 潘霄峰, 高云峰, 等. GSIG分布式光伏云缓存调度研究[J]. 中国测试, 2020, 46(6): 13-17
[5] 叶興, 薛家祥. 改进型LSTM网络光伏发电功率预测研究[J]. 中国测试, 2019, 45(11): 14-20
[6] BLAGA R, SABADUS A, STEFU N, et al. A current perspective on the accuracy of incoming solar energy forecasting[J]. Progress in Energy and Combustion Science, 2019, 70: 119-144
[7] 姚光辉. 光伏并网发电系统设计及MPPT技术研究[D]. 杭州: 浙江大学, 2014.
[8] MA T, YANG H, LU L. Solar photovoltaic system modeling and performance prediction[J]. Renewable and Sustainable Energy Reviews, 2014, 36: 304-315
[9] MAH A, NB B, KB C, et al. Ultra-short-term exogenous forecasting of photovoltaic power production using genetically optimized non-linear auto-regressive recurrent neural networks[J]. Renewable Energy, 2021, 171: 191-209
[10] 陈志宝, 丁杰, 周海, 等. 地基云图结合径向基函数人工神经网络的光伏功率超短期预测模型[J]. 中国电机工程学报, 2015, 35(3): 561-567
[11] WANG H, YI H, PENG J, et al. Deterministic and probabilistic forecasting of photovoltaic power based on deep convolutional neural network[J]. Energy Convers Manage, 2017, 153: 409-422
[12] WEN L, ZHOU K, YANG S, et al. Optimal load dispatch of community micro grid with deep learning based solar power and load forecasting[J]. Energy, 2019, 171(15): 1053-1065
[13] MASSAOUDI M, CHIHI I, SIDHOM L, et al. Enhanced random forest model for robust short-term photovoltaic power forecasting using weather measurements[J]. Energies, 2021, 14(13): 1-20
[14] HUANG C, KUO P. Multiple-input deep convolutional neural network model for short-term photovoltaic power forecasting[J]. IEEE Access, 2019, 7: 74822-74834
[15] ALZAHRANI A , SHAMSI P , DAGLI C, et al. Solar irradiance forecasting using deep neural networks[J]. Procedia Comput, 2017, 114: 304-313
[16] JIAN Z, YAN R, GAO R, et al. Performance enhancement of ensemble empirical mode decomposition[J]. Mechanical Systems & Signal Processing, 2010, 24(7): 2104-2123
[17] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780
