ULTRA-SHORT-TERM PHOTOVOLTAIC POWER FORECASTING CONSIDERING SIMILAR DAYS AND MB-T2V-KSA NETWORK

Li Yuhong; Bi Guihong; Yang Nan; Wang Xiaoling; Chen Shiyu

doi:10.19912/j.0254-0096.tynxb.2025-0342

PDF(7632 KB)

Acta Energiae Solaris Sinica ›› 2026, Vol. 47 ›› Issue (7) : 439-456. DOI: 10.19912/j.0254-0096.tynxb.2025-0342

ULTRA-SHORT-TERM PHOTOVOLTAIC POWER FORECASTING CONSIDERING SIMILAR DAYS AND MB-T2V-KSA NETWORK

Li Yuhong, Bi Guihong, Yang Nan, Wang Xiaoling, Chen Shiyu

Author information +

History +

Abstract

To address issues in existing photovoltaic power forecasting, such as redundant historical data, strong correlations in consecutive time series, large training datasets, slow model convergence, and the significant randomness and nonlinearity of power outputs, this paper proposes a novel forecasting approach based on improved grey relational analysis, dual-modal hybrid decomposition, and a multi-branch deep learning model. First, meteorological variables positively correlated with PV power output are selected through correlation analysis. Their positive correlation coefficients are incorporated as weights into an enhanced grey relational analysis framework, combined with a logarithmic normalization strategy to improve the accuracy and discriminative capability of similar-day sample selection. Next, a dual-modal hybrid decomposition is performed on the PV power time series using empirical wavelet transform （EWT） and swarm decomposition （SWD）, enabling multi-scale feature complementarity and generating more comprehensive input features. For model construction, a multi-branch input-parallel deep learning architecture, MB-T2V-KSA Net （Multi-Branch input-parallel Time2Vec-BiGRU-KAN-SA）, is developed. This model integrates Time2Vec encoding to capture both linear and periodic temporal features, employs a Bidirectional Gated Recurrent Unit （BiGRU） to learn contextual dependencies, utilizes the Kolmogorov-Arnold Network （KAN） for high-order feature interaction, and applies a Self-Attention （SA） mechanism to adaptively fuse multi-source features. Together, these components significantly enhance the model’s ability in temporal modeling and feature integration.Experimental results demonstrate that the proposed method achieves high prediction accuracy and robustness regardles of weather type classification and diverse typical weather conditions, with particularly strong performance in adverse weather scenarios.

Key words

PV power generation / similar days / grey relational analysis （GRA） / dual-mode hybrid decomposition / multi-branch architecture / deep learning / photovoltaic power forecasting

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

Li Yuhong, Bi Guihong, Yang Nan, Wang Xiaoling, Chen Shiyu. ULTRA-SHORT-TERM PHOTOVOLTAIC POWER FORECASTING CONSIDERING SIMILAR DAYS AND MB-T2V-KSA NETWORK[J]. Acta Energiae Solaris Sinica. 2026, 47(7): 439-456 https://doi.org/10.19912/j.0254-0096.tynxb.2025-0342

References

[1] 董存, 王铮, 白捷予, 等. 光伏发电功率超短期预测方法综述[J]. 高电压技术, 2023, 49(7): 2938-2951.
Dong C, Wang Z, Bai J Y, et al.Review of ultra-short-term forecasting methods for photovoltaic power generation[J]. High Voltage Engineering, 2023, 49(7): 2938-2951.
[2] 李超然, 潘鹏程, 杨伟荣, 等. 基于改进相似日优化HBA-BiLSTM-KELM的光伏发电功率预测[J]. 太阳能学报, 2024, 45(5): 508-516.
Li C R, Pan P C, Yang W R, et al.Research on PV system power prediction based on improvedsimilar dayand HBA-BiLSTM-KELM neural network[J]. Acta Energiae Solaris Sinica, 2024, 45(5): 508-516.
[3] 耿博, 高贞彦, 白恒远, 等. 结合相似日GA-BP神经网络的光伏发电预测[J]. 电力系统及其自动化学报, 2017, 29(6): 118-123.
Geng B, Gao Z Y, Bai H Y, et al.PV generation forecasting combined with similar days and GA-BP neural network[J]. Proceedings of the CSU-EPSA, 2017, 29(6): 118-123.
[4] 龚莺飞, 鲁宗相, 乔颖, 等. 光伏功率预测技术[J]. 电力系统自动化, 2016, 40(4): 140-151.
Gong Y F, Lu Z X, Qiao Y, et al.An overview of photovoltaic energy system output forecasting technology[J]. Automation of Electric Power Systems, 2016, 40(4): 140-151.
[5] 张宇华, 时鑫洋, 颜楠楠, 等. 逆向云灰色关联相似日的EEMD-RL-GWO-LSTM区域风光功率短期预测[J]. 太阳能学报, 2024, 45(10): 144-152.
Zhang Y H, Shi X Y, Yan N N, et al.Short-term prediction of regional wind-solar power of EEMD-RL-GWO-LSTM on reverse cloud grey correlation similar days[J]. Acta Energiae Solaris Sinica, 2024, 45(10): 144-152.
[6] 张大海, 孙锴, 和敬涵. 基于相似日与多模型融合的短期负荷预测[J]. 电网技术, 2023, 47(5): 1961-1969.
Zhang D H, Sun K, He J H.Short-term load forecasting based on similar day and multi-model fusion[J]. Power System Technology, 2023, 47(5): 1961-1969.
[7] 方朝雄, 郑洁云, 张章煌, 等. 基于相似日与VMD-DBO-KELM的分布式光伏发电功率预测方法[J]. 高电压技术, 2025, 51(7): 3477-3487.
Fang C X, Zheng J Y, Zhang Z H, et al.Distributed photovoltaic power prediction based on similar day and VMD-DBO-KELM[J]. High Voltage Engineering, 2025, 51(7): 3477-3487.
[8] 李争, 张杰, 徐若思, 等. 基于相似日聚类和PCC-VMD-SSA-KELM模型的短期光伏功率预测[J]. 太阳能学报, 2024, 45(2): 460-468.
Li Z, Zhang J, Xu R S, et al.Short term photovoltaic power prediction based on similar day clustering and PCC-VMD-SSA-KELM model[J]. Acta Energiae Solaris Sinica, 2024, 45(2): 460-468.
[9] 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 photovoltaic power prediction model based on comprehensive similar day selection[J]. Acta Energiae Solaris Sinica, 2025, 46(4): 301-312.
[10] 滕陈源, 丁逸超, 张有兵, 等. 基于VMD-Informer-BiLSTM模型的超短期光伏功率预测[J]. 高电压技术, 2023, 49(7): 2961-2971.
Teng C Y, Ding Y C, Zhang Y B, et al.Ultra-short-term photovoltaic power prediction based on VMD-Informer-BiLSTM model[J]. High Voltage Engineering, 2023, 49(7): 2961-2971.
[11] 孟鑫禹, 王睿涵, 张喜平, 等. 基于经验模态分解与多分支神经网络的超短期风功率预测[J]. 计算机应用, 2021, 41(1): 237-242.
Meng X Y, Wang R H, Zhang X P, et al.Ultra-short-term wind power prediction based on empirical mode decomposition and multi-branch neural network[J]. Journal of Computer Applications, 2021, 41(1): 237-242.
[12] 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.
[13] Zhang D D, Chen B A, Zhu H Y, et al.Short-term wind power prediction based on two-layer decomposition and BiTCN-BiLSTM-attention model[J]. Energy, 2023, 285: 128762.
[14] Kazemi S M, Goel R, Eghbali S, et al. Time2Vec: learning a vector representation of time[EB/OL]. arXiv:1907.05321. https://arxiv.org/abs/1907.05321.
[15] Ali M Fargalla M, Yan W, DENG J G, et al. TimeNet: Time2Vec attention-based CNN-BiGRU neural network for predicting production in shale and sandstone gas reservoirs[J]. Energy, 2024, 290: 130184.
[16] Geng D H, Wang B, Gao Q.A hybrid photovoltaic/wind power prediction model based on Time2Vec, WDCNN and BiLSTM[J]. Energy Conversion and Management, 2023, 291: 117342.
[17] Liu Z M, Wang Y X, Vaidya S, et al. KAN: Kolmogorov-Arnold networks[EB/OL]. arXiv:2404.19756. https://arxiv.org/abs/2404.19756.
[18] Sulaiman M H, Mustaffa Z, Mohamed A I, et al.Battery state of charge estimation for electric vehicle using Kolmogorov-Arnold networks[J]. Energy, 2024, 311: 133417.
[19] Ren D F, Hu Q, Zhang T D.Eklt: kolmogorov-arnold attention-driven LSTM with transformer model for river water level prediction[J]. Journal of Hydrology, 2025, 649: 132430.
[20] Qu J Q, Qian Z, Pei Y.Day-ahead hourly photovoltaic power forecasting using attention-based CNN-LSTM neural network embedded with multiple relevant and target variables prediction pattern[J]. Energy, 2021, 232: 120996.
[21] 艾雨, 贾燕冰, 韩肖清. 基于相似日筛选与组合深度学习模型的日前电价预测方法[J]. 电网技术, 2025, 49(1): 242-251.
Ai Y, Jia Y B, Han X Q.Day-ahead electricity price prediction method based on similar day screening and combined deep learning modeling[J]. Power System Technology, 2025, 49(1): 242-251.
[22] 张鹏, 齐波, 张若愚, 等. 基于经验小波变换和梯度提升径向基的变压器油中溶解气体预测方法[J]. 电网技术, 2021, 45(9): 3745-3754.
Zhang P, Qi B, Zhang R Y, et al.Dissolved gas prediction in transformer oil based on empirical wavelet transform and gradient boosting radial basis[J]. Power System Technology, 2021, 45(9): 3745-3754.
[23] Apostolidis G K, Hadjileontiadis L J.Swarm decomposition: a novel signal analysis using swarm intelligence[J]. Signal Processing, 2017, 132: 40-50.
[24] 张子豪, 黄悦华, 陈庆, 等. 基于VMD的CNN-BiGRU-ATT高海拔输电线路无线电干扰预测[J]. 高电压技术, 2025, 51(10): 5262-5273.
Zhang Z H, Huang Y H, Chen Q, et al.VMD based CNN-BiGRU-ATT high altitude radio interference prediction for power transmission lines[J]. High Voltage Engineering, 2025, 51(10): 5262-5273.
[25] Ahmad Khan Z, Hussain T, Baik S W.Dual stream network with attention mechanism for photovoltaic power forecasting[J]. Applied Energy, 2023, 338: 120916.
[26] 黄泽, 毕贵红, 谢旭, 等. 基于MBI-PBI-ResNet的超短期光伏功率预测[J]. 电力系统保护与控制, 2024, 52(2): 165-176.
Huang Z, Bi G H, Xie X, et al.Ultra-short-term PV power prediction based on MBI-PBI-ResNet[J]. Power System Protection and Control, 2024, 52(2): 165-176.
[27] 毕贵红, 张梓睿, 赵四洪, 等. 基于多模式分解和多分支输入的光伏功率超短期预测[J]. 高电压技术, 2024, 50(9): 3837-3849.
Bi G H, Zhang Z R, Zhao S H, et al.Ultra-short-term prediction of photovoltaic power based on multi-mode decomposition and multi-branch input[J]. High Voltage Engineering, 2024, 50(9): 3837-3849.
[28] 毕贵红, 赵鑫, 陈臣鹏, 等. 基于多通道输入和PCNN-BiLSTM的光伏发电功率超短期预测[J]. 电网技术, 2022, 46(9): 3463-3476.
Bi G H, Zhao X, Chen C P, et al.Ultra-short-term prediction of photovoltaic power generation based on multi-channel input and PCNN-BiLSTM[J]. Power System Technology, 2022, 46(9): 3463-3476.