融合中国气象局风能太阳能模式（CMA-WSP）输出的总辐射（GHI）和大气化学/沙尘模式（CMA-CUACE-DUST）的气溶胶预报,以及华北地区14个辐射观测站数据,集成共8种子模型与堆叠模型,构建面向气溶胶高影响天气的能源气象人工智能预报系统（AI-ENG）。该系统采用分月滑动的方法划分训练集与测试集,并在2023年和2024年春季沙尘天气高发的6个月期间,系统地应用与检验AI-ENG在华北地区的GHI预报性能。结果显示：1）研究时段内,华北地区沙尘天气频发且浓度偏高,CMA-WSP在强沙尘条件下常出现辐射高估,CMA-CUACE-DUST的预报误差与PM10浓度峰值呈正比。2）引入气溶胶预报产品后,AI-ENG系统可显著提升华北地区GHI预报的准确性。综合评估中极限梯度提升决策树（XGB）和多层感知器（MLP）表现最佳,7种性能检验指标均有不同程度的改善。3）在对2024年3月不同强度沙尘过程的日前72小时预报进行个例检验时,AI-ENG的预报误差下降17%~20%,尤其在平原和沙尘主传输通道区域效果更佳,但地形复杂区域仍存在一定的预报偏差。结果表明,AI-ENG系统能有效应用于沙尘天气太阳能短期预报。
Abstract
By integrating the global horizontal irradiance（GHI） forecast products from the China Meteorological Administration’s Wind and Solar Power Prediction System （CMA-WSP）, aerosol forecast products from the atmospheric chemical/dust model （CMA-CUACE-DUST）, and ground-based observations from 14 radiation observation stations in North China, an energy meteorology artificial intelligence forecasting system （AI-ENG） was developed for aerosol-high-impact weather. This system integrates a total of 8 sub-models and stacked generalization models. Employing a sliding monthly window strategy to partition the training and testing datasets, and the performance of AI-ENG in forecasting GHI was systematically evaluated over six months in the springs of 2023 and 2024. The results show that： 1） During the study period, dust weather events were frequent in North China with relatively high concentrations. CMA-WSP tended to overestimate radiation under strong dust conditions, and the forecast errors of CMA-CUACE-DUST were positively correlated with PM10 peak concentrations. 2） Incorporating aerosol forecast products, AI-ENG significantly improved the accuracy of GHI forecasts in North China. extreme gradient boosting decision tree（XGB） and multi-layer perceptron（MLP） models performed the best in the comprehensive evaluation, with all 7 metrics showing improvement. 3） In case studies of 72-hour day-ahead forecasts during different-intensity dust events in March 2024, AI-ENG reduced forecast errors by 17%-20%, particularly excelling in plain areas and along primary dust transport corridors. However, persistent biases remained in topographically complex regions. These results demonstrate that the AI-ENG system effectively supports short-term solar energy forecasting under dust weather conditions.
关键词
太阳辐射 /
预报方法 /
沙尘天气 /
数值模式 /
机器学习 /
气溶胶 /
雾-霾天气
Key words
solar radiation /
forecasting method /
dust weather /
numerical model /
machine learning /
aerosol /
haze-fog weather
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 国家能源局. 2025年1—11月份全国电力工业统计数据[EB/OL]. 2025[2025-12-26]. https://www.nea.gov.cn/20251226/640306962d7d421b921b902f48a04b47/c.html.
National Energy Administration. National Electric Power Industry Statistics(January-November2025)[EB/OL]. 2025 [2025-12-26]. https://www.nea.gov.cn/20251226/640306962d7d421b921b902f48a04b47/c.html
[2] 袁彬, 于廷照, 申彦波, 等. 基于人工智能多模式集成的光伏电站总辐射预报方法研究[J]. 太阳能学报, 2025, 46(4): 291-300.
Yuan B, Yu T Z, Shen Y B, et al.Research on artificial intelligence-based multi-model ensemble forecast of global horizontal irradiance in PV stations[J]. Acta Energiae Solaris Sinica, 2025, 46(4): 291-300.
[3] Chen Y, Chen S Y, Bi H R, et al.Where is the dust source of 2023 several severe dust events in China?[J]. Bulletin of the American Meteorological Society, 2024, 105(11): E2085-E2096.
[4] 申彦波, 赵宗慈, 石广玉. 地面太阳辐射的变化、影响因子及其可能的气候效应最新研究进展[J]. 地球科学进展, 2008, 23(9): 915-923.
Shen Y B, Zhao Z C, Shi G Y.The progress in variation of surface solar radiation, factors and probable climatic effects[J]. Advances in Earth Science, 2008, 23(9): 915-923.
[5] Breitkreuz H, Schroedter-Homscheidt M, Holzer-Popp T.A case study to prepare for the utilization of aerosol forecasts in solar energy industries[J]. Solar Energy, 2007, 81(11): 1377-1385.
[6] Kosmopoulos P G, Kazadzis S, Taylor M, et al.Dust impact on surface solar irradiance assessed with model simulations, satellite observations and ground-based measurements[J]. Atmospheric Measurement Techniques, 2017, 10(7): 2435-2453.
[7] Khalid H M, Rafique Z, Muyeen S M, et al.Dust accumulation and aggregation on PV panels: an integrated survey on impacts, mathematical models, cleaning mechanisms, and possible sustainable solution[J]. Solar Energy, 2023, 251: 261-285.
[8] 杨丽薇, 高晓清, 李振朝. 城市大气污染对到达地表的太阳辐射的影响研究[J]. 太阳能学报, 2022, 43(6): 124-131.
Yang L W, Gao X Q, Li Z C.Study on influence of urban air pollution on surface solar radiation[J]. Acta Energiae Solaris Sinica, 2022, 43(6): 124-131.
[9] Costa S C, Diniz A S A, Kazmerski L L. Solar energy dust and soiling R&D progress: literature review update for 2016[J]. Renewable and Sustainable Energy Reviews, 2018, 82(3): 2504-2536.
[10] 蔡子颖, 刘敬乐, 黄鹤, 等. 基于在线大气化学模式太阳能预报的改进[J]. 生态环境学报, 2016, 25(9): 1471-1475.
Cai Z Y, Liu J L, Huang H, et al.To improve simulation of solar radiation by atmospheric chemistry model WRF/CHEM in Tianjin[J]. Ecology and Environmental Sciences, 2016, 25(9): 1471-1475.
[11] Masoom A, Kosmopoulos P, Bansal A, et al.Forecasting dust impact on solar energy using remote sensing and modeling techniques[J]. Solar Energy, 2021, 228: 317-332.
[12] 商永朋, 郭军红, 宋煜, 等. 基于WRF-Solar模式和SCA订正的短期辐照度预报研究[J]. 太阳能学报, 2023, 44(5): 274-279.
Shang Y P, Guo J H, Song Y, et al.Research on short-term irradiance prediction based on WRF-Solar model and SCA correction[J]. Acta Energiae Solaris Sinica, 2023, 44(5): 274-279.
[13] Yang D Z.The future of solar forecasting in China[J]. Journal of Renewable and Sustainable Energy, 2023, 15(5): 052301.
[14] Gkikas A, Giannaros T M, Kotroni V, et al.Assessing the radiative impacts of an extreme desert dust outbreak and the potential improvements on short-term weather forecasts: the case of February 2015[J]. Atmospheric Research, 2019, 226: 152-170.
[15] Chuluunsaikhan T, Nasridinov A, Choi W S, et al.Predicting the power output of solar panels based on weather and air pollution features using machine learning[J]. Journal of Korea Multimedia Society, 2021, 24: 222-232.
[16] Lima F J L, Martins F R, Pereira E B, et al. Forecast for surface solar irradiance at the Brazilian Northeastern region using NWP model and artificial neural networks[J]. Renewable Energy, 2016, 87: 807-818.
[17] Chu Y H, Wang Y L, Yang D Z, et al.A review of distributed solar forecasting with remote sensing and deep learning[J]. Renewable and Sustainable Energy Reviews, 2024, 198: 114391.
[18] Fan J L, Wu L F, Ma X, et al.Hybrid support vector machines with heuristic algorithms for prediction of daily diffuse solar radiation in air-polluted regions[J]. Renewable Energy, 2020, 145: 2034-2045.
[19] Mo J Y, Gong S L, Zhang L, et al.Impacts of long-range transports from Central and South Asia on winter surface PM2.5 concentrations in China[J]. Science of the total environment, 2021, 777: 146243.
[20] Chen S Y, Zhao D, Huang J P, et al.Mongolia contributed more than 42% of the dust concentrations in Northern China in March and April 2023[J]. Advances in Atmospheric Sciences, 2023, 40(9): 1549-1557.
[21] 国家能源局. 2024年上半年光伏发电建设情况 [EB/OL]. 2024[2025-01-13]. https://www.nea.gov.cn/2024-07/25/c_1310782757.htm.National Energy Administration. Photovoltaic power generation construction in the first half of 2024 [EB/OL]. 2024[2025-01-13]. https://www.nea.gov.cn/2024-07/25/c_1310782757.htm.
[22] 中国气象局预报与网络司. 中国气象局风能太阳能气象预报系统(CMA_WSP)升级[EB/OL].2023[2025-02-13]. https://www.cma.gov.cn/2011xwzx/2011xqxkj/2011xkjdt/202305/t20230517_5510507.html.
China Meteorological Administration Forecast and Network Department. Upgrade of China Meteorological Administration Wind and Solar Weather Forecast System (CMA_WSP)[EB/OL].2023[2025-02-13]. https://www.cma.gov.cn/2011xwzx/2011xqxkj/2011xkjdt/202305/t20230517_5510507.html.
[23] Xia X A, Yang D Z, Shen Y B.Fengyun radiation services for solar energy meteorology: status and perspective[J]. Advances in Atmospheric Sciences, 2025, 42(2): 252-260.
[24] Gong S L, Zhang X Y, Zhao T L, et al. Characterization of soil dust aerosol in China and its transport and distribution during2001 ACE-Asia: 2. model simulation and validation[J]. Journal of Geophysical Research: Atmospheres, 2003, 108(D9): 2002JD002633.
[25] Gong S L, Zhang X Y.CUACE/Dust:an integrated system of observation and modeling systems for operational dust forecasting in Asia[J]. Atmospheric Chemistry and Physics, 2008, 8(9): 2333-2340.
[26] Kim S B, Yumimoto K, Uno I, et al.Dust model intercomparison between ADAM and CFORS/dust for Asian dust case in 2007 (March 28 - April 3)[J]. Sola, 2011, 7A: 25-28.
[27] Wang H, Xue M, Zhang X Y, et al.Mesoscale modeling study of the interactions between aerosols and PBL meteorology during a haze episode in Jing-Jin-Ji (China) and its nearby surrounding region-Part 1: aerosol distributions and meteorological features[J]. Atmospheric Chemistry and Physics, 2015, 15(6): 3257-3275.
[28] Ke H B, Gong S L, He J J, et al.Development and application of an automated air quality forecasting system based on machine learning[J]. Science of the Total Environment, 2022, 806: 151204.
[29] 环境保护部办公厅. 受沙尘天气过程影响城市空气质量评价补充规定 [EB/OL].2016. [2025-02-13]. https://www.mee.gov.cn/gkml/hbb/bgt/201701/t20170106_394054.htm.
General Office of the Ministry of Environmental Protection. Supplementary provisions on air quality assessment in cities affected by sand and dust weather processes [EB/OL].2016. [2025-02-13]. https://www.mee.gov.cn/gkml/hbb/bgt/201701/t20170106_394054.htm.
基金
中国气象局联合研究专项（24NLTSZ006）; 新疆“天池英才”引进计划（2023）; 中国气象局公共气象服务中心创新基金（M2024011）
PDF(7631 KB)
