{"title":"基于时空时序图像转换的光伏发电超短期预测","authors":"Md Tanjid Hossain, Yanfu Jiang, Xingyu Shi, Xutao Han, Zhiyi Li","doi":"10.1049/rpg2.70119","DOIUrl":null,"url":null,"abstract":"<p>Seasonal fluctuations and the intermittent nature of photovoltaic (PV) generation create significant challenges for accurate short-term forecasting. This study presents Next Frame Gramian Angular field U-Net (NFGUN), a hybrid deep learning forecasting framework that stands apart from conventional methods by transforming 1D PV time-series data into 2D Gramian Angular Summation Field (GASF) images. Unlike models that rely on direct regression or sky imagery, NFGUN forecasts the next GASF frame using a deep architecture and reconstructs it back into time-series form, effectively capturing nonlinear temporal dynamics. Its uniqueness lies in several key innovations: (1) the integration of Convolutional Long Short-Term Memory 2D (ConvLSTM2D) into a customised U-Net model for better generalisation spatiotemporal features; (2) the incorporation of residual blocks in the bottleneck to preserve deep features while mitigating vanishing gradients and cyclical encoding of time to enrich seasonal patterns; (3) the use of Lanczos interpolation with CIEDE2000 colour difference for high-precision reconstruction from predicted image frames. We evaluate NFGUN against six well-established forecasting methods and measure performance using six accuracy metrics such as MAE, RMSE, and WAPE across all four seasons; NFGUN demonstrates superior performance. Compared to the best-performing benchmark, it achieved improvements in MAE (61.23% winter, 56% spring, 37.45% summer, 59.67% autumn), RMSE (48.34% winter, 64.63% spring, 31.65% summer, 45.83% autumn), and WAPE (49.9% winter, 43.84% spring, 45.83% summer, 48.72% autumn), underscoring its ability to adapt to seasonal variability. These results demonstrate NFGUN's ability to effectively capture complex, seasonal dynamics, making it a robust solution for ultra-short-term PV power forecasting.</p>","PeriodicalId":55000,"journal":{"name":"IET Renewable Power Generation","volume":"19 1","pages":""},"PeriodicalIF":2.9000,"publicationDate":"2025-08-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/rpg2.70119","citationCount":"0","resultStr":"{\"title\":\"Ultra-Short-Term Forecasting of Photovoltaic Power Generation through Spatiotemporal Time-Series Image Conversion\",\"authors\":\"Md Tanjid Hossain, Yanfu Jiang, Xingyu Shi, Xutao Han, Zhiyi Li\",\"doi\":\"10.1049/rpg2.70119\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p>Seasonal fluctuations and the intermittent nature of photovoltaic (PV) generation create significant challenges for accurate short-term forecasting. This study presents Next Frame Gramian Angular field U-Net (NFGUN), a hybrid deep learning forecasting framework that stands apart from conventional methods by transforming 1D PV time-series data into 2D Gramian Angular Summation Field (GASF) images. Unlike models that rely on direct regression or sky imagery, NFGUN forecasts the next GASF frame using a deep architecture and reconstructs it back into time-series form, effectively capturing nonlinear temporal dynamics. Its uniqueness lies in several key innovations: (1) the integration of Convolutional Long Short-Term Memory 2D (ConvLSTM2D) into a customised U-Net model for better generalisation spatiotemporal features; (2) the incorporation of residual blocks in the bottleneck to preserve deep features while mitigating vanishing gradients and cyclical encoding of time to enrich seasonal patterns; (3) the use of Lanczos interpolation with CIEDE2000 colour difference for high-precision reconstruction from predicted image frames. We evaluate NFGUN against six well-established forecasting methods and measure performance using six accuracy metrics such as MAE, RMSE, and WAPE across all four seasons; NFGUN demonstrates superior performance. Compared to the best-performing benchmark, it achieved improvements in MAE (61.23% winter, 56% spring, 37.45% summer, 59.67% autumn), RMSE (48.34% winter, 64.63% spring, 31.65% summer, 45.83% autumn), and WAPE (49.9% winter, 43.84% spring, 45.83% summer, 48.72% autumn), underscoring its ability to adapt to seasonal variability. These results demonstrate NFGUN's ability to effectively capture complex, seasonal dynamics, making it a robust solution for ultra-short-term PV power forecasting.</p>\",\"PeriodicalId\":55000,\"journal\":{\"name\":\"IET Renewable Power Generation\",\"volume\":\"19 1\",\"pages\":\"\"},\"PeriodicalIF\":2.9000,\"publicationDate\":\"2025-08-14\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/rpg2.70119\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"IET Renewable Power Generation\",\"FirstCategoryId\":\"5\",\"ListUrlMain\":\"https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/rpg2.70119\",\"RegionNum\":4,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"IET Renewable Power Generation","FirstCategoryId":"5","ListUrlMain":"https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/rpg2.70119","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
引用次数: 0
摘要
季节波动和光伏发电的间歇性为准确的短期预测带来了重大挑战。本研究提出了Next Frame Gramian Angular field U-Net (NFGUN),这是一种混合深度学习预测框架,通过将1D PV时间序列数据转换为2D Gramian Angular sum field (GASF)图像,与传统方法不同。与依赖直接回归或天空图像的模型不同,NFGUN使用深度架构预测下一个GASF帧,并将其重建为时间序列形式,有效地捕获非线性时间动态。它的独特性在于几个关键的创新:(1)将卷积长短期记忆2D (ConvLSTM2D)集成到定制的U-Net模型中,以更好地概括时空特征;(2)在瓶颈处加入残差块,保留深度特征,同时减轻梯度消失和时间周期编码,丰富季节模式;(3)利用CIEDE2000色差的Lanczos插值对预测图像帧进行高精度重建。我们根据六种成熟的预测方法评估NFGUN,并使用六个精度指标(如MAE, RMSE和WAPE)在所有四个季节测量性能;NFGUN表现出优越的性能。与表现最好的基准相比,其MAE(冬季61.23%,春季56%,夏季37.45%,秋季59.67%)、RMSE(冬季48.34%,春季64.63%,夏季31.65%,秋季45.83%)和WAPE(冬季49.9%,春季43.84%,夏季45.83%,秋季48.72%)均有所改善,体现了其适应季节变化的能力。这些结果表明,NFGUN能够有效捕获复杂的季节性动态,使其成为超短期光伏发电预测的强大解决方案。
Ultra-Short-Term Forecasting of Photovoltaic Power Generation through Spatiotemporal Time-Series Image Conversion
Seasonal fluctuations and the intermittent nature of photovoltaic (PV) generation create significant challenges for accurate short-term forecasting. This study presents Next Frame Gramian Angular field U-Net (NFGUN), a hybrid deep learning forecasting framework that stands apart from conventional methods by transforming 1D PV time-series data into 2D Gramian Angular Summation Field (GASF) images. Unlike models that rely on direct regression or sky imagery, NFGUN forecasts the next GASF frame using a deep architecture and reconstructs it back into time-series form, effectively capturing nonlinear temporal dynamics. Its uniqueness lies in several key innovations: (1) the integration of Convolutional Long Short-Term Memory 2D (ConvLSTM2D) into a customised U-Net model for better generalisation spatiotemporal features; (2) the incorporation of residual blocks in the bottleneck to preserve deep features while mitigating vanishing gradients and cyclical encoding of time to enrich seasonal patterns; (3) the use of Lanczos interpolation with CIEDE2000 colour difference for high-precision reconstruction from predicted image frames. We evaluate NFGUN against six well-established forecasting methods and measure performance using six accuracy metrics such as MAE, RMSE, and WAPE across all four seasons; NFGUN demonstrates superior performance. Compared to the best-performing benchmark, it achieved improvements in MAE (61.23% winter, 56% spring, 37.45% summer, 59.67% autumn), RMSE (48.34% winter, 64.63% spring, 31.65% summer, 45.83% autumn), and WAPE (49.9% winter, 43.84% spring, 45.83% summer, 48.72% autumn), underscoring its ability to adapt to seasonal variability. These results demonstrate NFGUN's ability to effectively capture complex, seasonal dynamics, making it a robust solution for ultra-short-term PV power forecasting.
期刊介绍:
IET Renewable Power Generation (RPG) brings together the topics of renewable energy technology, power generation and systems integration, with techno-economic issues. All renewable energy generation technologies are within the scope of the journal.
Specific technology areas covered by the journal include:
Wind power technology and systems
Photovoltaics
Solar thermal power generation
Geothermal energy
Fuel cells
Wave power
Marine current energy
Biomass conversion and power generation
What differentiates RPG from technology specific journals is a concern with power generation and how the characteristics of the different renewable sources affect electrical power conversion, including power electronic design, integration in to power systems, and techno-economic issues. Other technologies that have a direct role in sustainable power generation such as fuel cells and energy storage are also covered, as are system control approaches such as demand side management, which facilitate the integration of renewable sources into power systems, both large and small.
The journal provides a forum for the presentation of new research, development and applications of renewable power generation. Demonstrations and experimentally based research are particularly valued, and modelling studies should as far as possible be validated so as to give confidence that the models are representative of real-world behavior. Research that explores issues where the characteristics of the renewable energy source and their control impact on the power conversion is welcome. Papers covering the wider areas of power system control and operation, including scheduling and protection that are central to the challenge of renewable power integration are particularly encouraged.
The journal is technology focused covering design, demonstration, modelling and analysis, but papers covering techno-economic issues are also of interest. Papers presenting new modelling and theory are welcome but this must be relevant to real power systems and power generation. Most papers are expected to include significant novelty of approach or application that has general applicability, and where appropriate include experimental results. Critical reviews of relevant topics are also invited and these would be expected to be comprehensive and fully referenced.
Current Special Issue. Call for papers:
Power Quality and Protection in Renewable Energy Systems and Microgrids - https://digital-library.theiet.org/files/IET_RPG_CFP_PQPRESM.pdf
Energy and Rail/Road Transportation Integrated Development - https://digital-library.theiet.org/files/IET_RPG_CFP_ERTID.pdf