地球科学进展 ›› 2026, Vol. 41 ›› Issue (1): 102 -112. doi: 10.11867/j.issn.1001-8166.2026.011

新型能源系统中气象技术的创新与应用专栏 上一篇    下一篇

盘古模型在一个海上典型风电场功率预测中的应用研究
邓华1(), 王晨宇2, 张艳霞1   
  1. 1.中国气象局广州热带海洋气象研究所/粤港澳大湾区气象研究院,广东 广州 510640
    2.国家超级计算无锡中心,江苏 无锡 214072
  • 收稿日期:2025-09-12 修回日期:2025-11-20 出版日期:2026-01-10
  • 基金资助:
    粤港澳大湾区气象科技协同攻关项目(GHMA2024Z01)

A Study on the Application of the Pangu Model in Power Prediction for a Typical Offshore Wind Farm

Hua Deng1(), Chenyu Wang2, Yanxia Zhang1   

  1. 1.Guangzhou Institute of Tropical and Marine Meteorology China Meteorological Administration / Guangdong-Hong Kong-Macao Greater Bay Area Academy of Meteorological Research, Guangzhou 510640, China
    2.National Supercomputing Center in Wuxi, Wuxi Jiangsu 214072, China
  • Received:2025-09-12 Revised:2025-11-20 Online:2026-01-10 Published:2026-03-10
  • About author:Deng Hua, research areas include numerical weather prediction models for wind and solar power generation, power prediction, and data assimilation. E-mail: hdeng@gd121.cn
  • Supported by:
    the Guangdong-Hong Kong-Macao Greater Bay Area Collaborative Project on Key Meteorological Technologies(GHMA2024Z01)

随着风电功率预测对时效和更新频率的要求不断提高,基于数值天气预报进行预测的方法局限性愈发明显,人工智能气象模型的发展为风电功率预测提供了高效的气象预报来源。选取广东阳江一典型海上风电场作为服务对象,应用盘古模型和数值天气预报模式CMA-GD进行2023年和2024年共17个月的模拟计算,设计了4组功率预测试验,对比分析了不同气象数据预报源、不同功率预测方法以及进行风速订正对于功率预测的改进效果。试验结果显示,在风电功率预测中盘古模型具备同等甚至超越数值天气预报模型的精准度。应用生成的场站功率曲线做功率预测时,基于盘古模型气象预报的短期功率预测准确率超过基于CMA-GD的准确率,甚至在2024年个别月份的预测效果与应用机器学习算法预测的功率相当。根据2023年11月的试验结果,基于订正的盘古模型预报和应用ResMLP算法的功率预测模型,功率预测效果最佳,24 h预测准确率为0.714,96 h预测准确率为0.654,将该模型应用到2024年1~6月,效果稳定,可以达到电网的考核标准。研究为在风电功率预测业务中应用盘古模型或其他人工智能气象模型提供了一种可借鉴的方案。

With increasing demands for timeliness and update frequency in wind power prediction, the limitations of methods based on Numerical Weather Prediction (NWP) have become increasingly evident. The development of artificial intelligence meteorological models has provided an efficient source of meteorological forecasts for wind power prediction. Selecting an offshore wind farm in Yangjiang, Guangdong as the service object, the Pangu AI meteorological model and the numerical weather prediction model CMA-GD were applied to conduct simulation calculations for a total of 17 months in 2023 and 2024. Four sets of power prediction experiments were designed, which used different meteorological data forecast sources, different power prediction methods, and a wind speed correction method, to compare and analyze the improvement of wind power prediction. The experimental results show that the Pangu model achieves accuracy comparable to or even surpassing that of numerical weather prediction models. When using the generated site power curve for power prediction, the short-term power prediction accuracy based on the Pangu model meteorological forecast exceeds that based on CMA-GD, and even in some months of 2024, the prediction performance is comparable to that of using machine learning algorithms for power prediction. According to the November 2023 experimental results, the power prediction model based on corrected Pangu model forecasts and the ResMLP algorithm achieved the best power prediction performance, with a 24-hour prediction accuracy of 0.714 and a 96-hour prediction accuracy of 0.654. Applying this model to the period from January to June 2024 yielded stable results that met the grid assessment standards. This study provides a referable solution for applying the Pangu or other artificial intelligence meteorological models in wind power prediction operations.

中图分类号: 

图1 20231~11月风电功率随风速变化散点图
Fig. 1 Scatter plot of wind power versus wind speed from January to November in 2023
图2 ResMLP模型结构示意图
Fig. 2 Schematic diagram of machine learning model ResMLP structure
表1 ResMLP模型超参数配置
Table 1 Hyperparameter configuration for machine learning models ResMLP
表2 风电功率预测试验方案
Table 2 Test scheme for wind power prediction
图3 202311月观测风速(横坐标)和预报风速(纵坐标)散点图
(a)CMA-GD模式预报;(b)盘古模型预报;黑线为线性拟合结果。
Fig. 3 The scatter plot of observedhorizontal coordinatevs. forecastedvertical coordinatewind speed for November in 2023
(a) Forecast from CMA-GD; (b) Forecast from PanGu; The black line is the result of linear fit.
图4 202311月实际功率(横坐标)和预报功率(纵坐标)散点图
(a)基于CMA-GD风电功率预报;(b)基于盘古模型风电功率预报;黑线为线性拟合结果。
Fig. 4 The scatter plot of realhorizontal coordinatevs. forecastedvertical coordinatewind power for November in 2023
(a) Forecast from CMA-GD; (b) Forecast from PanGu; The black line is the result of linear fit.
图5 202311月逐日预测的未来逐小时风电功率准确度
(a)预报0~24 h; (b)预报24~48 h; (c)预报48~72 h; (d)预报72~96 h。
Fig. 5 Hourly accuracy of wind power predictions for November in 2023
(a)0~24 h forecast; (b) 24~48 h forecast;(c) 48~72 h forecast;(d) 72~96 h forecast.
图6 202311月风速预测均方根误差
Fig. 6 Root mean square error for wind forecasting for November in 2023
图7 202311月逐日预报未来逐小时风电功率与实际风电功率
(a)预报0~24 h;(b)预报24~48 h;(c)预报48~72 h;(d)预报72~96 h;阴影表示在实际风电功率上下浮动40%的误差范围。
Fig. 7 Forecasted and actual wind power over time for November in 2023
(a)0~24 h forecast; (b) 24~48 h forecast;(c) 48~72 h forecast;(d) 72~96 h forecast; Shaded area indicating ±40% error margin relative to actual wind power.
图8 20241~6月日功率预测准确率的分布箱式图
Fig. 8 Box plot of the distribution of day power prediction accuracy from January to June in 2024
图9 20241~6Test-4不同预报时效的日功率预测准确率
Fig. 9 Daily cumulative power forecasting accuracy with different leading time from January to June in 2024
[1] Zhang Yachao, Liu Kaipei, Liao Xiaobing, et al. Multi-time scale source-load coordination dispatch model for power system with large-scale wind power[J]. High Voltage Engineering201945(2): 600-608.
张亚超, 刘开培, 廖小兵, 等. 含大规模风电的电力系统多时间尺度源荷协调调度模型研究[J]. 高电压技术201945(2): 600-608.
[2] Yetis Y, Tehrani K, Jamshidi M. Wind speed forecasting using machine learning approach based on meteorological data—a case study[J]. Energy and Environment Research202212(2): 11.
[3] Hui Xiaojian, Wang Zhen, Zhang Shanwen, et al. Wind power forecast based on ARMA[J]. Modern Electronics Technique201639(7): 145-148, 153.
惠小健, 王震, 张善文, 等. 基于ARMA的风电功率预测[J]. 现代电子技术201639(7): 145-148, 153.
[4] Li Li, Ye Lin. Short-term wind power forecasting based on an improved persistence approach[J]. Transactions of the Chinese Society of Agricultural Engineering201026(12): 182-187.
李丽, 叶林. 基于改进持续法的短期风电功率预测[J]. 农业工程学报201026(12): 182-187.
[5] Wang S X, Shi J R, Yang W, et al. High and low frequency wind power prediction based on transformer and BiGRU-Attention[J]. Energy2024288: 129753.
[6] Sun Rongfu, Zhang Tao, He Qing, et al. Review on key technologies and applications in wind power forecasting[J]. High Voltage Engineering202147(4): 1 129-1 143.
孙荣富, 张涛, 和青, 等. 风电功率预测关键技术及应用综述[J]. 高电压技术202147(4): 1 129-1 143.
[7] Zhang Taihao, Yin Feng, Yuan Ping. Overview of wind speed and wind power forecasting methods[J]. Modern Computer2021(8): 45-48.
张泰豪, 殷锋, 袁平. 风速及风功率预测方法综述[J]. 现代计算机2021(8): 45-48.
[8] Zhang Z W, Lin L L, Gao S, et al. A machine learning model for hub-height short-term wind speed prediction[J]. Nature Communications202516: 3195.
[9] Guo X J, Zeng P L, Xiong X, et al. Short-term wind power forecasting methods based on machine learning: a review and case study[J]. Energy Reports202514: 3 753-3 782.
[10] Selvaraj J, Muthuramalingam L, Karthikeyan V, et al. Optimizing wind energy integration: a review of forecasting techniques and emerging trends[J]. Archives of Computational Methods in Engineering2025. DOI: 10.1007/s11831-025-10442-1 .
[11] Wang Lei, Teng Wei, Wu Xin, et al. A review of key technologies and research applications in wind power prediction[J/OL]. Journal of Chinese Society of Power Engineering2025. [2025-08-10]. .
王磊, 滕伟, 武鑫, 等. 风功率预测关键技术及其研究应用综述[J/OL]. 动力工程学报2025. [2025-08-10]. .
[12] Yang M, Han C, Zhang W, et al. A short-term power prediction method based on numerical weather prediction correction and the fusion of adaptive spatiotemporal graph feature information for wind farm cluster[J]. Expert Systems with Applications2025274: 126979.
[13] Chen Ying, Zhou Hai, Wang Wenpeng, et al. Improvement of ultra-short-term forecast for wind power[J]. Automation of Electric Power Systems201135(15): 30-33, 87.
陈颖, 周海, 王文鹏, 等. 风电场输出功率超短期预测结果分析与改进[J]. 电力系统自动化201135(15): 30-33, 87.
[14] Ye Lin, Liu Peng. Combined model based on EMD-SVM for short-term wind power prediction[J]. Proceedings of the CSEE201131(31): 102-108.
叶林, 刘鹏. 基于经验模态分解和支持向量机的短期风电功率组合预测模型[J]. 中国电机工程学报201131(31): 102-108.
[15] Liao Xuechao, Wu Jieping, Chen Caisheng. Short-term wind power prediction model combining attention mechanism and LSTM[J]. Computer Engineering202248(9): 286-297, 304.
廖雪超, 伍杰平, 陈才圣. 结合注意力机制与LSTM的短期风电功率预测模型[J]. 计算机工程202248(9): 286-297, 304.
[16] Wang Yuxuan, Liu Erjia, Huang Yongzhang. An ultra-short-term wind power prediction method based on CNN-LSTM-lightGBM combination[J]. Science Technology and Engineering202222(36): 16 067-16 074.
王愈轩, 刘尔佳, 黄永章. 基于CNN-LSTM-lightGBM组合的超短期风电功率预测方法[J]. 科学技术与工程202222(36): 16 067-16 074.
[17] Gao Sheng, Xu Peihua, Chen Zhenghong. Wind speed multi-mode ensemble forecasting for wind farms based on machine learning[J]. Southern Energy Construction202411(1): 85-95.
高盛, 许沛华, 陈正洪. 基于机器学习的风电场风速多模式集合预报[J]. 南方能源建设202411(1): 85-95.
[18] Zhang Luna, Feng Qiang, Liu Liqun, et al. Application of ensemble selection method in short-term wind power forecasting[J]. Journal of Tropical Meteorology202440(2): 285-296.
张路娜, 冯强, 刘立群, 等. 集合选优方法在短期风功率预测中的应用研究[J]. 热带气象学报202440(2): 285-296.
[19] Yang M, Guo Y F, Huang T, et al. Power prediction considering NWP wind speed error tolerability: a strategy to improve the accuracy of short-term wind power prediction under wind speed offset scenarios[J]. Applied Energy2025377: 124720.
[20] Yang M, Guo Y F, Fan F L, et al. Two-stage correction prediction of wind power based on numerical weather prediction wind speed superposition correction and improved clustering[J]. Energy2024302: 131797.
[21] Lam R, Sanchez-Gonzalez A, Willson M, et al. Learning skillful medium-range global weather forecasting[J]. Science2023382(6 677): 1 416-1 421.
[22] Chen L, Zhong X H, Zhang F, et al. FuXi: a cascade machine learning forecasting system for 15-day global weather forecast[J]. NPJ Climate and Atmospheric Science20236: 190.
[23] Bodnar C, Bruinsma W P, Lucic A, et al. A foundation model for the Earth system[J]. Nature2025641(8 065): 1 180-1 187.
[24] Pathak J, Subramanian S, Harrington P, et al. FourCastNet: a global data-driven high-resolution weather model using adaptive Fourier neural operators[PP]. V1. arXiv (2022-02-22)[2025-12-20]. DOI: 10.48550/arXiv.2202.11214 .
[25] Bi K F, Xie L X, Zhang H H, et al. Accurate medium-range global weather forecasting with 3D neural networks[J]. Nature2023619(7 970): 533-538.
[26] Shen Xueshun, Wang Jianjie, Li Zechun, et al. China’s independent and innovative development of numerical weather prediction[J]. Acta Meteorologica Sinica202078(3): 451-476.
沈学顺, 王建捷, 李泽椿, 等. 中国数值天气预报的自主创新发展[J]. 气象学报202078(3): 451-476.
[27] Hubei Meteorological Service Center (Hubei Professional Meteorological Service Station). Wind power abnormal data processing method based on sigmoid curve and dual-envelope algorithm:CN117235449A[P]. 2023-12-15.
湖北省气象服务中心(湖北省专业气象服务台). 基于Sigmoid 曲线及双包络算法的风电功率异常数据处理方法:CN117235449A[P]. 2023-12-15.
[28] Park J Y, Lee J K, Oh K Y, et al. Development of a novel power curve monitoring method for wind turbines and its field tests[J]. IEEE Transactions on Energy Conversion201429(1): 119-128.
[29] Cao Lixin, Liu Weimin, Guo Huquan. Cleaning and modeling of abnormal data of wind farm power curve[J]. Journal of Lanzhou University of Technology202248(4): 64-70.
曹立新, 刘伟民, 郭虎全. 风电场功率曲线异常数据的清洗与建模[J]. 兰州理工大学学报202248(4): 64-70.
[30] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA. IEEE, 2016: 770-778.
[1] 胡琳曼, 宋晓蕾, 尹志聪. 梅雨极端化的表征和预测[J]. 地球科学进展, 2026, 41(1): 40-48.
[2] 潘林林, 靳双龙, 宋宗朋, 丁煌, 胡睿, 肖子牛, 杜杰, 杨静, 包庆, 王勃. 风能和太阳能短中期气象预报技术及其最新进展[J]. 地球科学进展, 2026, 41(1): 73-86.
[3] 唐学远. 中国极地冰冻圈人工智能技术应用的进展与展望[J]. 地球科学进展, 2025, 40(8): 778-794.
[4] 刘定洲, 刘虹辰, 张锦昌, 张江阳, 周鹏程, 郭鑫, 李璐, 黄彦铭. 基于U-Net卷积神经网络的海底磁异常条带自动识别方法研究——以沙茨基海隆为例[J]. 地球科学进展, 2025, 40(8): 864-876.
[5] 张井勇, 杨占梅, 吴凌云. 夏季极端高温预测模型系统及实际应用[J]. 地球科学进展, 2025, 40(5): 516-524.
[6] 黄建平, 连鑫博, 王睿, 王丹凤, 黄忠伟, 张北斗, 胡淑娟. 大气病原微生物监测预警的研究进展[J]. 地球科学进展, 2025, 40(4): 331-347.
[7] 周杰, 巨能攀, 张燕, 田华兵, 何朝阳. 泥石流视频图像跟踪检测方法研究[J]. 地球科学进展, 2025, 40(4): 388-400.
[8] 王凡, 张旭东, 任沂斌, 刘颖洁, 王浩宇, 李晓峰. 基于人工智能技术的海洋智能预报研究进展[J]. 地球科学进展, 2025, 40(2): 111-125.
[9] 朱学明, 王海龙, 郭绍敬, 张旭日. 南海中尺度涡旋预报研究进展[J]. 地球科学进展, 2025, 40(11): 1112-1128.
[10] 张井勇. 面向碳中和的气候变化预估与风险研究新框架[J]. 地球科学进展, 2025, 40(1): 15-20.
[11] 吴涛, 徐泽阳, 闫文琦, 费李莹, 刘荷冲, 赵靖舟, 李军, 杜治伟. 准噶尔盆地莫索湾地区侏罗系超压预测技术研究[J]. 地球科学进展, 2024, 39(4): 429-439.
[12] 李珺, 赵杨, 陈钊州, 张乐乐, 曹欢, 李世昌. 基于地球物理驱动的裂缝性地层三维坍塌压力预测及应用[J]. 地球科学进展, 2024, 39(11): 1196-1209.
[13] 贾黎黎, 李婷婷, 朱鑫, 易隆科, 罗思亮. 基于改进随机森林模型的多要素层次插值技术在土壤硒元素空间分布上的运用[J]. 地球科学进展, 2023, 38(12): 1259-1270.
[14] 王劲松, 姚玉璧, 王莺, 王素萍, 刘晓云, 周悦, 杜昊霖, 张宇, 任余龙. 青藏高原地区气象干旱研究进展与展望[J]. 地球科学进展, 2022, 37(5): 441-461.
[15] 摆玉龙, 路亚妮, 刘名得. 基于变分模态分解的机器学习模型择优风速预测系统[J]. 地球科学进展, 2021, 36(9): 937-949.
阅读次数
全文


摘要