基于时频域分解-增强-融合Transformer的多因素光伏发电功率预测

罗强; 高崇; 曹华珍; 何璇; 曾庆彬; 张勇军

doi:10.19912/j.0254-0096.tynxb.2024-1912

PDF(1666 KB)

太阳能学报 ›› 2026, Vol. 47 ›› Issue (3) : 604-615. DOI: 10.19912/j.0254-0096.tynxb.2024-1912

基于时频域分解-增强-融合Transformer的多因素光伏发电功率预测

罗强¹, 高崇¹, 曹华珍¹, 何璇¹, 曾庆彬^2,3, 张勇军³

作者信息 +

MULTI-FACTOR PHOTOVOLTAIC POWER GENERATION PREDICTION BASED ON TIME-FREQUENCY DOMAIN DECOMPOSITION ENHANCEMENT FUSION TRANSFORMER

Luo Qiang¹, Gao Chong¹, Cao Huazhen¹, He Xuan¹, Zeng Qingbin^2,3, Zhang Yongjun³

Author information +

文章历史 +

摘要

针对现有光伏发电功率预测方法对于光伏发电功率变化曲线趋势性和随机性挖掘不足导致性能受限的问题,提出基于时频域分解-增强-融合Transformer的多因素光伏发电功率预测方法。首先,构建编码器-解码器模型,利用时域增强模块和频域增强模块,将光伏发电功率的历史和当前数据分别分解为周期项和趋势项;再通过时域和频域增强注意力模块挖掘当前数据和历史数据的语义关系。其次,采用随机层获取当前数据的随机项。然后,通过多级自适应的方式融合光伏发电功率数据的周期项、趋势项和随机项用于预测。此外,协同考虑太阳辐射、湿度、风速和温度等多因素,采用多变量通道独立的方式,降低数据冗余,进一步提升模型性能。仿真结果说明相比于其他方法,所提模型能够有效提高光伏发电功率的预测精度。

Abstract

A multi-factor photovoltaic power prediction method based on a time-frequency domain decomposition enhancement fusion Transformer is proposed to address the performance limitations caused by insufficient mining of the trend and randomness of the photovoltaic power change curve in existing photovoltaic power prediction methods. Firstly, an encoder-decoder framework is constructed, in which historical and current PV power data are decomposed into periodic and trend components through time-domain and frequency-domain enhancement modules. The semantic relationships between current and historical data are further extracted using time-domain and frequency-domain enhanced attention mechanisms. Subsequently, a stochastic layer is employed to capture the stochastic component of the current data. Then, a multi-level adaptive fusion strategy integrates the periodic, trend, and stochastic components of PV power data for forecasting. In addition, multiple factors such as solar irradiance, humidity, wind speed, and temperature are incorporated using an independent multi-channel feature extraction approach to reduce data redundancy and further improve model performance. Experiments on PVGD-1 and DKASC-Sanyo datasets show that the proposed method outperforms Informer, Autoformer, and Fedformer.

导出引用

罗强, 高崇, 曹华珍, 何璇, 曾庆彬, 张勇军. 基于时频域分解-增强-融合Transformer的多因素光伏发电功率预测[J]. 太阳能学报. 2026, 47(3): 604-615 https://doi.org/10.19912/j.0254-0096.tynxb.2024-1912

Luo Qiang, Gao Chong, Cao Huazhen, He Xuan, Zeng Qingbin, Zhang Yongjun. MULTI-FACTOR PHOTOVOLTAIC POWER GENERATION PREDICTION BASED ON TIME-FREQUENCY DOMAIN DECOMPOSITION ENHANCEMENT FUSION TRANSFORMER[J]. Acta Energiae Solaris Sinica. 2026, 47(3): 604-615 https://doi.org/10.19912/j.0254-0096.tynxb.2024-1912

中图分类号： TM615

参考文献

[1] 谭显东, 刘俊, 徐志成, 等. “双碳”目标下“十四五”电力供需形势[J]. 中国电力, 2021, 54(5): 1-6.
TAN X D, LIU J, XU Z C, et al.Power supply and demand balance during the 14th five-year plan period under the goal of carbon emission peak and carbon neutrality[J]. Electric power, 2021, 54(5): 1-6.
[2] 李翌为. 分布式光伏并网对配电网电压的影响[J]. 自动化应用, 2024(5): 103-105.
LI Y W.Impact of distributed photovoltaic grid connection on distribution network voltage[J]. Automation application, 2024(5): 103-105.
[3] 徐其春, 李良杰, 卢泽汉. 电网接入光伏发电功率的实时调度与预测研究[J]. 电网与清洁能源, 2023, 39(12): 141-147.
XU Q C, LI L J, LU Z H.A study on the real-time scheduling and prediction of photovoltaic power connected to the power grid[J]. Power system and clean energy, 2023, 39(12): 141-147.
[4] 赵滨滨, 王莹, 王彬, 等. 基于ARIMA时间序列的分布式光伏系统输出功率预测方法研究[J]. 可再生能源, 2019, 37(6): 820-823.
ZHAO B B, WANG Y, WANG B, et al.Photovoltaic power prediction in distribution network based on ARIMA model time series[J]. Renewable energy resources, 2019, 37(6): 820-823.
[5] 刘大贵, 王维庆, 张慧娥, 等. 基于隐马尔科夫修正的光伏中长期电量预测及调度计划应用[J]. 高电压技术, 2023, 49(2): 840-848.
LIU D G, WANG W Q, ZHANG H E, et al.Mid-long term available quantity of electricity forecasting with error calibration by hidden markov model in photovoltaic and application of dispatching plan[J]. High voltage engineering, 2023, 49(2): 840-848.
[6] 熊川羽, 廖晓红, 何诗英, 等. 使用快速傅里叶变换优化周期参数的EMD-FFT-SARIMA光伏发电预测模型[J]. 强激光与粒子束, 2024, 36(8): 121-127.
XIONG C Y, LIAO X H, HE S Y, et al.EMD-FFT-SARIMA photovoltaic power generation prediction model using fast fourier transform optimization cycle parameters[J]. High power laser and particle beams, 2024, 36(8): 121-127.
[7] 罗琦, 杨俊华, 黄逸, 等. 基于变分模式分解和向量自回归模型的波浪发电系统输出功率预测[J]. 太阳能学报, 2023, 44(3): 291-297.
LUO Q, YANG J H, HUANG Y, et al.Output power prediction of wave power system based on variational model decomposition and vector autoregressive model[J]. Acta energiae solaris sinica, 2023, 44(3): 291-297.
[8] 叶进, 卢泉, 王钰淞, 等. 基于级联随机森林的光伏故障诊断模型研究[J]. 太阳能学报, 2021, 42(3): 358-362.
YE J, LU Q, WANG Y S, et al.Research on PV fault diagnosis model based on cascaded random forest[J]. Acta energiae solaris sinica, 2021, 42(3): 358-362.
[9] 刘志超, 袁三男, 唐万成. 基于BLSTM-随机森林的短期光伏发电输出功率预测[J]. 电源技术, 2021, 45(4): 495-498.
LIU Z C, YUAN S N, TANG W C.Output power prediction of short-term photovoltaic power generation based on BLSTM-random forest[J]. Chinese journal of power sources, 2021, 45(4): 495-498.
[10] 陈海宏, 易永利, 黄珅, 等. 基于CatBoost算法的短期光伏功率预测方法[J]. 浙江电力, 2023, 42(2): 67-75.
CHEN H H, YI Y L, HUANG Q, et al.Research on a short-term photovoltaic power prediction method based on CatBoost[J]. Zhejiang electric power, 2023, 42(2): 67-75.
[11] 徐恒山, 莫汝乔, 薛飞, 等. 基于时间戳特征提取和CatBoost-LSTM模型的光伏短期发电功率预测[J]. 太阳能学报, 2024, 45(5): 565-575.
XU H S, MO R Q, XUE F, et al.Short-term photovoltaic power prediction based on timestamp feature extration and CatBoost-LSTM model[J]. Acta energiae solaris sinica, 2024, 45(5): 565-575.
[12] ZHOU B W, CHEN X Y, LI G D, et al.XGBoost-SFS and double nested stacking ensemble model for photovoltaic power forecasting under variable weather conditions[J]. Sustainability, 2023, 15(17): 13146.
[13] 李永飞, 张耀, 林帆, 等. 基于气候特征分析及改进XGBoost算法的中长期光伏电站发电量预测方法[J]. 电力系统保护与控制, 2024, 52(11): 84-92.
LI Y F, ZHANG Y, LIN F, et al.Medium-and long-term power generation forecast based on climate characterization and an improved XGBoost algorithm for photovoltaic power plants[J]. Power system protection and control, 2024, 52(11): 84-92.
[14] 张程珂, 刘会灯, 朱渝宁, 等. 基于多特征分析提取的随机森林超短期光伏功率预测[J]. 电力需求侧管理, 2023, 25(6): 50-56.
ZHANG C K, LIU H D, ZHU Y N, et al.Ultra-short-term photovoltaic power prediction for random forests based on multiple feature analysis and extraction[J]. Power demand side management, 2023, 25(6): 50-56.
[15] 刘源延, 孔小兵, 马乐乐, 等. 基于小波包变换与深度学习的超短期光伏功率预测[J]. 太阳能学报, 2024, 45(5): 537-546.
LIU Y Y, KONG X B, MA L L, et al.Ultra-short-term photovoltaic power forecasting based on wavelet packet transform and deep learning[J]. Acta energiae solaris sinica, 2024, 45(5): 537-546.
[16] 彭曙蓉, 郑国栋, 黄士峻, 等. 基于XGBoost算法融合多特征短期光伏发电量预测[J]. 电测与仪表, 2020, 57(24): 76-83.
PENG S R, ZHENG G D, HUANG S J, et al.Multiple-feature short-term photovoltaic generation forecasting based on XGBoost algorithm[J]. Electrical measurement & instrumentation, 2020, 57(24): 76-83.
[17] 范国庆, 李康辉, 高捷, 等. 基于核密度估计和CatBoost算法的光伏功率预测方法[J]. 上海电力大学学报, 2023, 39(6): 529-535.
FAN G Q, LI K H, GAO J, et al.A photovoltaic power prediction method based on kernel density estimation and CatBoost method[J]. Journal of Shanghai University of Electric Power, 2023, 39(6): 529-535.
[18] TANG P W, ZHANG Q H, ZHANG X C.A recurrent neural network based generative adversarial network for long multivariate time series forecasting[C]//Proceedings of the 2023 ACM International Conference on Multimedia Retrieval. Thessaloniki, Greece, 2023: 181-189.
[19] HOSSAIN M S, MAHMOOD H.Short-term photovoltaic power forecasting using an LSTM neural network and synthetic weather forecast[J]. IEEE access, 2020, 8: 172524-172533.
[20] ZHOU X, PANG C X, ZENG X H, et al.A short-term power prediction method based on temporal convolutional network in virtual power plant photovoltaic system[J]. IEEE transactions on instrumentation and measurement, 2023, (76): 1-10.
[21] WAN R Z, MEI S P, WANG J, et al.Multi-variate temporal convolutional network: a deep neural networks approach for multivariate time series forecasting[J]. Electronics, 2019, 8(8): 876.
[22] SHAMSOLMOALI P, ZAREAPOOR M, GRANGER E, et al.Setformer is what you need for vision and language[J]. Proceedings of the AAAI conference on artificial intelligence, 2024, 38(5): 4713-4721.
[23] WANG W H, XIE E Z, LI X, et al.Pyramid vision transformer: a versatile backbone for dense prediction without convolutions[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Montreal, Canada, 2021: 568-578.
[24] 王慧强, 陈楚皓, 吕宏武, 等. 基于双向稀疏Transformer的多变量时序分类模型[J]. 小型微型计算机系统, 2024, 45(3): 555-561.
WANG H Q, CHEN C H, LU H W, et al.Multivariate time series classification model based on bidirectional sparse Transformer[J]. Journal of Chinese computer systems, 2024, 45(3): 555-561.
[25] 黄莉, 甘恒玉, 刘兴举, 等. 基于Transformer编码器的超短期光伏发电功率预测[J]. 智慧电力, 2024, 52(5): 16-22, 59.
HUANG L, GAN H Y, LIU X J, et al.Ultra-short-term photovoltaic power generation prediction based on Transformer encoder[J]. Smart power, 2024, 52(5): 16-22, 59.
[26] ZHOU H Y, ZHANG S H, PENG J Q, et al.Informer: Beyond efficient transformer for long sequence time-series forecasting[J]. Proceedings of the AAAI conference on artificial intelligence, 2021, 35(12): 11106-11115.
[27] LIU S Z, YU H, LIAO C, et al.Pyraformer: low-complexity pyramidal attention for long-range time series modeling and forecasting[C]//International Conference on Learning Representations. 2022.
[28] NIE Y Q, NGUYEN Nan H, SINTHONG P, et al.A time series is worth 64 words: long-term forecasting with transformers[C]//International Conference on Learning Representations. 2023.
[29] WU H, XU J H, WANG J M, et al.Autoformer: decomposition transformers with auto-correlation for long-term series forecasting[C]//Advances in Neural Information Processing Systems. 2021.
[30] ZHOU T, MA Z, WEN Q S, et al.Fedformer: frequency enhanced decomposed transformer for long-term series forecasting[C]//International Conference on Machine Learning. 2022.