基于综合相似日选取和二次分解的NGO-CNN-LSTM光伏功率短期预测研究

宋煜; 许野; 王旭; 李薇; 陈哲

doi:10.19912/j.0254-0096.tynxb.2024-1188

PDF(2392 KB)

太阳能学报 ›› 2025, Vol. 46 ›› Issue (11) : 219-234. DOI: 10.19912/j.0254-0096.tynxb.2024-1188

基于综合相似日选取和二次分解的NGO-CNN-LSTM光伏功率短期预测研究

宋煜, 许野, 王旭, 李薇, 陈哲

作者信息 +

SHORT-TERM PREDICTION OF PV POWER OF NGO-CNN-LSTM BASED ON COMPREHENSIVE SIMILAR DAY SELECTION AND QUADRATIC DECOMPOSITION

Song Yu, Xu Ye, Wang Xu, Li Wei, Chen Zhe

Author information +

文章历史 +

摘要

针对当前光伏发电功率预测的数据预处理过程中相似日选取评估标准单一、原始功率序列呈现波动性和非平稳性以及预测模型的最佳参数组合生成存在困难等问题,提出一种基于相似日选取和二次分解的北方苍鹰优化卷积神经网络-长短期记忆网络（NGO-CNN-LSTM）日前光伏发电功率短期预测模型。首先,在利用皮尔逊相关系数法选取主要气象因素的基础上,运用欧氏距离和弗雷歇距离相结合的综合相似日选取法选定待预测日的历史相似日,构成训练集合;其次,运用完全自适应噪声集合经验模态分解（CEEMDAN）对训练集的历史功率序列进行一次模态分解,并引入多尺度排列熵（MPE）对高熵值分量进行重构;然后,使用基于北方苍鹰算法（NGO）的变分模态分解（VMD）方法对重构的非平稳分量进行二次分解,并根据多尺度排列熵（MPE）值将两次分解的分量进行重构;最后,分别构建重构序列的NGO-CNN-LSTM模型,并将预测结果进行叠加得到最终预测结果。通过云南光伏电站案例分析表明,基于相似日选取和二次分解的NGO-CNN-LSTM模型的预测精度更高,能对电站生产规划调度的合理制定提供科学有效的指导。

Abstract

Aiming at some problems involved into data preprocessing process of PV power prediction such as single evaluation criteria for similar day selection, fluctuating and non-stationary characteristics of original power series, as well as the difficulty in generating optimal parameter combination of the prediction model, this paper proposes a day-ahead NGO-CNN-LSTM （northern goshawk optimization-convolutional neural network-long short-term memory） short-term PV output prediction model based on similar day selection and double decomposition. Firstly, the Pearson correlation coefficient is used to identify the main meteorological factors; then, the comprehensive similar day selection method combining Euclidean distance and Frechet distance is utilized to select the historical similar day for the predicted day and generate the training set. Secondly, complete ensemble empirical mode decomposition with adaptive noise （CEEMDAN） is used to decompose the historical power sequence of the training set; correspondingly, multi-scale permutation entropy（MPE） is introduced to reconstruct the components with high entropy values. Next, variational mode decomposition （VMD） based on northern goshawk optimization （NGO） is used for the secondary decomposition of reconstructed non-stationary components. The components are reconstructed according to their MPE values. Finally, NGO-CNN-LSTM models corresponding to each reconstructed sequence are constructed, where their predicted results are summed to obtained the final prediction results. The application of the proposed model in PV power station of Yunnan province shows that compared with other benchmark models, NGO-CNN-LSTM model based on similar day selection and double decomposition has the higher prediction accuracy, which has the practical and feasible guide significance for the production plan of power station and the reasonable formulation of electricity market participation strategy.

导出引用

宋煜, 许野, 王旭, 李薇, 陈哲. 基于综合相似日选取和二次分解的NGO-CNN-LSTM光伏功率短期预测研究[J]. 太阳能学报. 2025, 46(11): 219-234 https://doi.org/10.19912/j.0254-0096.tynxb.2024-1188

Song Yu, Xu Ye, Wang Xu, Li Wei, Chen Zhe. SHORT-TERM PREDICTION OF PV POWER OF NGO-CNN-LSTM BASED ON COMPREHENSIVE SIMILAR DAY SELECTION AND QUADRATIC DECOMPOSITION[J]. Acta Energiae Solaris Sinica. 2025, 46(11): 219-234 https://doi.org/10.19912/j.0254-0096.tynxb.2024-1188

中图分类号： TM615

参考文献

[1] 葛乐, 陆文伟, 袁晓冬, 等. 基于改进相似日和ABC-SVM的光伏电站功率预测[J]. 太阳能学报, 2018, 39(3): 775-782.
GE L, LU W W, YUAN X D, et al.Power forecasting of photovoltaic plant based on improved similar day and ABC-SVM[J]. Acta energiae solaris sinica, 2018, 39(3): 775-782.
[2] 王璐. 去年全球可再生能源新增装机中国贡献过半[N]. 经济参考报, 2024-01-26(2).
WANG L. Last year, China contributed more than half of the world's newly installed renewable energy capacity[N]. Economic reference report, 2024-01-26(2).
[3] 蔡源, 吴浩, 唐丹. 光伏发电功率预测方法综述[J]. 四川电力技术, 2024, 47(2): 25-31.
CAI Y, WU H, TANG D.Reviews of photovoltaic power prediction methods[J]. Sichuan electric power technology, 2024, 47(2): 25-31.
[4] 郭雯博. 基于相似日和Tent-SSA-BP光伏发电功率预测研究[J]. 信息技术与信息化, 2023(5): 173-177.
GUO W B.Research on photovoltaic power prediction based on similar day and Tent-SSA-BP[J]. Information technology and informatization, 2023(5): 173-177.
[5] 乔路丽, 方诗琦, 赵庭锐, 等. 基于相似日和IGA-BP的光伏发电功率预测方法研究[J]. 电网与清洁能源, 2022, 38(1): 128-134.
QIAO L L, FANG S Q, ZHAO T R, et al.A study on the forecasting method of photovoltaic power generation based on similar day and IGA-BP[J]. Power system and clean energy, 2022, 38(1): 128-134.
[6] 赵一鉴, 林利, 王茜蒨, 等. 基于大地距离计算相似度的海上目标轨迹预测[J]. 计算机应用, 2023, 43(11): 3594-3598.
ZHAO Y J, LIN L, WANG Q Q, et al.Trajectory prediction of sea targets based on geodetic distance similarity calculation[J]. Journal of computer applications, 2023, 43(11): 3594-3598.
[7] 李宏仲, 叶翔宇, 付国. 信号分解技术在新能源发电功率预测中的应用评述[J]. 南方电网技术, 2023, 17(4): 3-15.
LI H Z, YE X Y, FU G.Review on the application of signal decomposition technology in power prediction of new energy power generation[J]. Southern power system technology, 2023, 17(4): 3-15.
[8] 陈龙, 张菁, 张昊立, 等. 基于VMD和射箭算法优化改进ELM的短期光伏发电预测[J]. 太阳能学报, 2023, 44(10): 135-141.
CHEN L, ZHANG J, ZHANG H L, et al.Short-term photovoltaic power generation forecast based on VMD-IAA-IHEKLM model[J]. Acta energiae solaris sinica, 2023, 44(10): 135-141.
[9] 曹哲, 赵葵银, 王田宇, 等. 基于EMD-AO-DELM的光伏功率预测算法[J]. 湖南工程学院学报(自然科学版), 2023, 33(4): 6-14.
CAO Z, ZHAO K Y, WANG T Y, et al.EMD-AO-DELM based PV power prediction algorithm[J]. Journal of Hunan Institute of Engineering (natural science edition), 2023, 33(4): 6-14.
[10] 董雪, 赵宏伟, 赵生校, 等. 基于SOM聚类和二次分解的BiGRU超短期光伏功率预测[J]. 太阳能学报, 2022, 43(11): 85-93.
DONG X, ZHAO H W, ZHAO S X, et al.Ultra-short-term forecasting method of photovoltaic power based on SOM clustering, secondary decomposition and BiGRU[J]. Acta energiae solaris sinica, 2022, 43(11): 85-93.
[11] 吉兴全, 赵国航, 叶平峰, 等. 基于QMD-HBi GRU的短期光伏功率预测方法[J]. 高电压技术, 2024, 50(9): 3850-3859.
JI X Q, ZHAO G H, YE P F, et al.Short-term PV forecasting method based on the QMD-HBi GRU model[J]. High voltage engineering, 2024, 50(9): 3850-3859.
[12] LIN W S, ZHANG B, LI H Y, et al.Multi-step prediction of photovoltaic power based on two-stage decomposition and BiLSTM[J]. Neurocomputing, 2022, 504: 56-67.
[13] 李瑞, 范玉刚. 基于CEEMDAN多尺度排列熵和SO-RELM的高压隔膜泵单向阀故障诊断[J]. 振动与冲击, 2023, 42(5): 127-135.
LI R, FAN Y G.Fault diagnosis of one-way valve of high-pressure diaphragm pump based on CEEMDAN multi-scale permutation entropy and SO-RELM[J]. Journal of vibration and shock, 2023, 42(5): 127-135.
[14] 李翰章, 冯江涛, 王鹏程, 等. 基于TDE-SO-AWM-GRU的光伏发电功率预测模型[J]. 中国电力, 2024, 57(12): 41-49.
LI H Z, FENG J T, WANG P C, et al.Photovoltaic power prediction model based on TDE-SO-AWM-GRU[J]. Electric power, 2024, 57(12): 41-49.
[15] 黄牧涛, 邢芳菲, 陈兴邦, 等. 基于K-means聚类和极限学习机组合算法的短期光伏功率预测[J]. 水电能源科学, 2024, 42(2): 217-220, 216.
HUANG M T, XING F F, CHEN X B, et al.Short-term PV power prediction based on K-means clustering and extreme learning machine combination algorithm[J]. Water resources and power, 2024, 42(2): 217-220, 216.
[16] LI S, YANG J, WU F Z, et al.Combined prediction of photovoltaic power based on sparrow search algorithm optimized convolution long and short-term memory hybrid neural network[J]. Electronics, 2022, 11(10): 1654.
[17] 杨锡运, 王诗晨, 张艳峰, 等. 基于相似日的Grey-Markov与BP_Adaboost的短期光伏功率预测[J]. 电源技术, 2023, 47(6): 790-794.
YANG X Y, WANG S C, ZHANG Y F, et al.Short-term PV power prediction based on Grey-Markov and BP_Adaboost by similar days[J]. Chinese journal of power sources, 2023, 47(6): 790-794.
[18] 张文昊. 弗雷歇距离判断曲线相似度的嵌入式模块[J]. 单片机与嵌入式系统应用, 2020, 20(9): 17-20, 50.
ZHANG W H.Embedded module of curve similarity judging based on Fréchet distance[J]. Microcontrollers & embedded systems, 2020, 20(9): 17-20, 50.
[19] 宋煜, 许野, 刘锋平, 等. 基于综合相似日选取的SO-CNN-LSTM光伏功率预测模型研究[J]. 太阳能学报, 2025, 46(4): 301-312.
SONG Y, XU Y, LIU F P, et al.Research on SO-CNN-LSTM photocoltaic power prediction model based on comprehensive similar day selection[J]. Acta energiae solaris sinica, 2025,46(4): 301-312.
[20] 李芬, 孙凌, 王亚维, 等. 基于CEEMDAN-GSA-LSTM和SVR的光伏功率短期区间预测[J]. 上海交通大学学报, 2024, 58(6): 806-818.
LI F, SUN L, WANG Y W, et al.Short-term interval forecasting of photovoltaic power based on CEEMDAN-GSA-LSTM and SVR[J]. Journal of Shanghai Jiao Tong University, 2024, 58(6): 806-818.
[21] 王粟, 江鑫, 曾亮, 等. 基于VMD-DESN-MSGP模型的超短期光伏功率预测[J]. 电网技术, 2020, 44(3): 917-926.
WANG S, JIANG X, ZENG L, et al.Ultra-short-term photovoltaic power prediction based on VMD-DESN-MSGP model[J]. Power system technology, 2020, 44(3): 917-926.
[22] 蒲维, 杨毅强, 张渊博, 等. 基于NGO-VMD-FCBF-Informer的电力负荷组合预测模型[J]. 智能计算机与应用, 2023, 13(11): 135-141.
PU W, YANG Y Q, ZHANG Y B, et al.Powerload forecasting based on NGO-VMD-FCBF-Informer[J]. Intelligent computer and applications, 2023, 13(11): 135-141.
[23] 张琳, 高胜强, 宋煜, 等. 基于ALIF-VMD二次分解的NGO-CNN-LSTM电力负荷短期组合预测模型[J]. 科学技术与工程, 2025, 25(11): 4583-4597.
ZHANG L, GAO S Q, SONG Y, et al.NGO-CNN-LSTM power load Short-term combination forecasting model based on ALIF-VMD quadratic decomposition[J]. Science technology and engineering, 2025, 25(11): 4583-4597.
[24] 胡以怀, 李从跃, 沈威, 等. 基于VMD-多尺度排列熵和SVM的船用空压机故障诊断方法[J]. 中国测试, 2024, 50(6): 20-27.
HU Y H, LI C Y, SHEN W, et al.Fault diagnosis of marine air compressor based on VMD multi-scale permutation entropy and SVM[J]. China measurement & test, 2024, 50(6): 20-27.
[25] 郭宇辰, 加鹤萍, 余涛, 等. 基于CNN-LSTM组合模型的碳价预测方法[J]. 科技管理研究, 2023, 43(11): 200-206.
GUO Y C, JIA H P, YU T, et al.Carbon price prediction method based on CNN-LSTM combination model[J]. Science and technology management research, 2023, 43(11): 200-206.
[26] 张子豪. 基于AM-CNN-LSTM光伏短期功率预测研究[D]. 沈阳: 沈阳农业大学, 2023.
ZHANG Z H.Research on photovoltaic short-term power prediction based on AM-CNN-LSTM[D]. Shenyang: Shenyang Agricultural University, 2023.
[27] 刘旭丽, 莫毓昌, 吴哲, 等. 基于DWT-CNN-LSTM的超短期光伏发电功率预测[J]. 郑州大学学报(理学版), 2022, 54(4): 86-94.
LIU X L, MO Y C, WU Z, et al.Super-short-term photovoltaic power forecasting based on DWT-CNN-LSTM[J]. Journal of Zhengzhou University (natural science edition), 2022, 54(4): 86-94.
[28] 张子华, 李琰, 徐天奇, 等. 基于麻雀算法优化的VMD-CNN-LSTM的短期风电功率研究[J]. 电气传动, 2023, 53(5): 77-83.
ZHANG Z H, LI Y, XU T Q, et al.Research on short-term wind power forecasting based on VMD-CNN-LSTM optimized by sparrow algorithm[J]. Electric drive, 2023, 53(5): 77-83.