Study on the characteristics of solar shortwave irradiance and comparative analysis of short-term irradiance prediction of Yangbajing, Tibet
-
摘要:
本研究利用西藏羊八井太阳短波辐照度观测数据分析了该地区2020—2021年的辐射时间序列分布特征，基于时间序列分析、随机森林（Random Forest，RF）和Prophet进行建模预测，通过对比研究探究三种模型在该地区的适用性以及提高模型预测精度的方法.结果表明：该地区短波太阳辐照度呈"双峰"倒"U"型分布的月变化和"单峰"倒"U"型分布的日变化特征.RF在选用模型中最优，其标准化均方根误差（Normalized Root Mean Square Error，NRMSE）、决定系数R2分别为17.54%和0.962.小波变换去噪能提高各模型预测精度，NRMSE降低4.82%~12.94%.组合模型能提高预测精度，误差倒数权重组合模型的NRMSE较差分自回归滑动平均模型（Autoregressive Integrated Moving Average，ARIMA）和Prophet分别下降35.22%、25.12%.预测时间步长差异也会影响预测效果，模型的预测误差随时间步长逐渐增大而减小.因此，可利用RF等机器学习模型在西藏地区进行太阳辐照度短期预测，通过小波变换、组合模型、预测时间步长等环节提高预测精度，以满足当地光伏发电对太阳辐照度的预测需求.
Abstract:In this research, using the measured solar short-wave irradiance data during the year of 2020 and 2021 at Yangbajing observation station in Tibet, the characteristics of radiation time series distribution is analyzed. And three solar irradiance prediction models tailor to Yangbajing area are established based on time series analysis, random forest (RF) and Prophet. Moreover, by comparing this three models, the applicability of three models and the method for improving the prediction accuracy of three models are explored. Our result shows that the monthly and diurnal variation of short-wave solar irradiance in this area display a "bimodal" inverted "U" and a "unipolar" inverted "U" distribution, respectively. Among the three models, RF is found to be the best model for predicting the solar irradiance in this area, with NRMSE (Normalized Root Mean Square Error) and R2 of 17.54% and 0.962, respectively. Both wavelet transform denoising and combination model can improve the prediction accuracy of the three models and the NRMSE by applying wavelet transform denoising is reduced by 4.82%~12.94%.The NRMSE of autoregressive integrated moving average model (ARIMA) and Prophet of the error reciprocal weight combination model decreased by 35.22% and 25.12%, respectively. Furthermore, prediction time step differences also affect the prediction effect, and the prediction error of the model gradually becomes smaller with the time step. Therefore, machine learning models such as RF can be used to predict solar irradiance in Tibet, and the prediction accuracy can be improved through wavelet transforms, combined models, prediction time steps, etc., in order to meet the forecasting needs of local photovoltaic power generation for solar irradiance.
-
Key words:
- Solar irradiance /
- Short-term forecasting /
- ARIMA /
- Random forest (RF) /
- Prophet /
- Yangbajing
-
-
图 1
羊八井地区太阳短波辐照度时序图(单位：W·m-2)
Figure 1.
Sequence diagram of solar shortwave irradiance in Yangbajing area (unit: W·m-2)
图 2
太阳短波辐照度时序拟合模型残差序列自相关、偏自相关图
Figure 2.
Autocorrelation and partial autocorrelation graphs of residual sequence of solar short wave irradiance time series fitting model
图 3
羊八井地区太阳短波辐照度时序小波分解图(单位：W·m-2)
Figure 3.
Temporal wavelet decomposition of solar shortwave irradiance in Yangbajing area (unit: W·m-2)
图 4
羊八井地区太阳短波辐照度箱线图(单位：W·m-2)
Figure 4.
Box plot of solar irradiance in Yangbajing area (unit: W·m-2)
图 5
羊八井地区太阳短波辐照度实测值与预测值箱线图(单位：W·m-2)
Figure 5.
Box plot of measured versus predicted solar short-wave irradiance in Yangbajing area (unit: W·m-2)
图 6
羊八井地区太阳短波辐照度时序模型预测值与实测值散点图(单位：W·m-2)
Figure 6.
Scatter plot of predicted versus measured solar irradiance sequence model in Yangbajing area (unit: W·m-2)
表 1
羊八井地区太阳短波辐照度时序ADF检验结果
Table 1.
Sequential ADF test results of solar shortwave irradiance in Yangbajing area
检验统计值 P值 检验统计量临界值
(α=0.01)检验统计量临界值
(α=0.05)检验统计量临界值
(α=0.10)-9.836 0.007 -3.431 -2.862 -2.567 
下载: 导出CSV
表 2
羊八井地区太阳短波辐照度原始时间序列、残差时间序列纯随机性检验结果
Table 2.
Pure randomness test results of original time series and residual time series of solar short-wave irradiance in Yangbajing area
延迟阶数 原始序列纯随机性检验结果 拟合模型残差序列纯随机性检验结果 LB检验统计量 P值 LB检验统计量 P值 6 20179.611 0.00001 4.672 0.587 12 37931.543 0.00011 19.179 0.084
下载: 导出CSV
表 3
随机森林模型参数
Table 3.
Parameters of random forest model
最大特征数 弱学习器个数 最大深度 节点划分最小样本数 叶子节点最小样本数 13 13 22 2 1
下载: 导出CSV
表 4
Prophet模型参数
Table 4.
Parameters of prophet model
突变点范围 周期项灵活度参数 趋势项灵活度参数 突变点数量 年周期季节项 0.8 30 0.05 70 20
下载: 导出CSV
表 5
2020年、2021年羊八井地区太阳短波辐照度月变化
Table 5.
Monthly variation of solar shortwave irradiance in Yangbajing area during 2020 and 2021
2020年(W·m-2) 2021年(W·m-2) 最大值 平均值 标准差 最大值 平均值 标准差 1月 863.0 323.1 279.5 859.5 363.5 294.0 2月 1004.1 408.1 326.1 1054.3 428.3 330.4 3月 1233.0 450.9 345.4 1106.9 466.4 336.1 4月 1220.2 479.1 349.2 1266.1 486.3 334.2 5月 1264.1 491.1 339.1 1174.4 483.1 341.4 6月 1235.5 544.3 346.0 1226.8 586.3 364.8 7月 1233.2 489.3 346.9 1248.7 495.9 365.8 8月 1253.4 507.9 361.2 1283.5 435.9 352.0 9月 1121.7 467.0 353.4 1245.4 442.9 335.7 10月 1053.6 473.8 336.2 1000.5 437.7 323.8 11月 931.3 371.3 298.0 887.4 391.4 307.8 12月 761.4 327.1 276.0 929.0 332.8 278.7
下载: 导出CSV
表 6
2020年、2021年羊八井地区太阳短波辐照度日变化
Table 6.
The daily variation of shortwave solar irradiation in Yangbajing area during 2020 and 2021
2020年(W·m-2) 2021年(W·m-2) 最大值 平均值 标准差 最大值 平均值 标准差 7时 66.0 8.6 13.1 58.8 9.2 14.7 8时 311.6 77.2 78.0 311.9 76.2 80.1 9时 604.7 252.8 133.6 542.9 251.9 135.6 10时 792.0 476.9 153.0 781.5 474.4 154.5 11时 1029.5 662.2 171.8 1006.1 671.3 170.8 12时 1178.7 784.4 202.2 1226.8 793.1 192.0 13时 1248.0 826.3 232.6 1249.1 838.5 221.1 14时 1264.1 795.1 243.3 1283.5 797.8 245.6 15时 1187.0 690.2 252.4 1177.0 690.1 242.5 16时 1095.4 558.9 226.1 1035.6 570.1 213.7 17时 823.2 386.3 188.5 812.8 381.2 176.2 18时 630.9 198.3 133.4 553.3 199.3 129.4 19时 324.1 59.5 68.4 279.6 60.9 64.2
下载: 导出CSV
表 7
羊八井地区太阳短波辐照度时序模型预测值和实测值之间误差分析
Table 7.
Analysis of errors between predicted and measured solar irradiance time series model in Yangbajing area
模型 小波变换去噪处理的模型 权重组合模型 ARIMA RF Prophet WT-ARIMA WT-RF WT-Prophet CP-average CP-mape CP-variance RMSE/(W·m-2) 125.2 53.9 108.3 109 51.3 97.6 89.1 81.1 88.9 NRMSE/% 40.75 17.54 35.24 34.98 16.87 31.82 29.0 26.4 28.95 R2 0.825 0.962 0.854 0.818 0.961 0.850 0.900 0.917 0.905
下载: 导出CSV
表 8
不同预测时间步长太阳短波辐照度时序模型预测值与实测值之间误差分析
Table 8.
Analysis of errors between predicted value and measured value of solar short wave irradiance with different sequence model
ARIMA RF Prophet RMSE
(W·m-2)NRMSE
(%)RMSE
(W·m-2)NRMSE
(%)RMSE
(W·m-2)NRMSE
(%)1 h 125.2 40.75 53.9 17.54 108.3 34.98 2 h 116.5 35.02 51.5 15.48 101.9 30.63 3 h 114.8 34.51 50.3 15.11 98.8 29.70
下载: 导出CSV
-
Benali L, Notton G, Fouilloy A, et al. 2019. Solar radiation forecasting using artificial neural network and random forest methods: Application to normal beam, horizontal diffuse and global components. Renewable Energy, 132: 871-884, doi: 10.1016/j.renene.2018.08.044.
Benmouiza K, Cheknane A. 2016. Small-scale solar radiation forecasting using ARMA and nonlinear autoregressive neural network models. Theoretical and Applied Climatology, 124(3-4): 945-958, doi: 10.1007/s00704-015-1469-z.
Breiman L. 2001. Random Forests. Machine Learning, 45: 5-32, doi: 10.1023/A:1010933404324.
Cohen A. 1994. Ten Lectures on Wavelets, CBMS-NSF regional conference series in applied mathematics, Vol. 61, Ⅰ. Daubechies, SIAM, xix+357 pp. Journal of Approximation Theory, 78(3): 460-461, doi: 10.1006/jath.1994.1093.
Ferrari S, Fina A, Lazzaroni M, et al. 2012. Illuminance prediction through statistical models. //2012 IEEE Workshop on Environmental Energy and Structural Monitoring Systems. Perugia, Italy: IEEE, 90-96, doi:
10.1109/eesms.2012.6348406 .Hao Y Z, Li X H, Hu Y N, et al. 2021. Change rules and influencing factors of 57 years of solar energy resources in Inner Mongolia. Acta Energiae Solaris Sinica (in Chinese), 42(9): 145-151, doi: 10.19912/j.0254-0096.tynxb.2019-0868.
Hou N, Zhang X T, Wei Y, et al. 2021. Estimation of surface incident shortwave radiation over China based on random forest regression method. Acta Energiae Solaris Sinica (in Chinese), 42(2): 31-36, doi: 10.19912/j.0254-0096.tynxb.2018-0989.
Huang J H, Liu H. 2021. A hybrid decomposition-boosting model for short-term multi-step solar radiation forecasting with NARX neural network. Journal of Central South University, 28(2): 507-526, doi: 10.1007/s11771-021-4618-9.
Ji W, Chee K C. 2011. Prediction of hourly solar radiation using a novel hybrid model of ARMA and TDNN. Solar Energy, 85(5): 808-817, doi: 10.1016/j.solener.2011.01.013.
King J C. 1996. Longwave atmospheric radiation over Antarctica. Antarctic Science, 8(1): 105-109. doi: 10.1017/S0954102096000132
Li Y, Sheng Z, Jing J R. 2019. Feature analysis of stratospheric wind and temperature fields over the Antigua site by rocket data. Earth and Planetary Physics, 3: 414-424, doi: 10.26464/epp2019040.
Li Z, Zhang G, Li D, et al. 2007. Application and development of solar energy in building industry and its prospects in China. Energy Policy, 35: 4121-4127, doi: 10.1016/j.enpol.2007.02.006.
Liu J, Cao M Y, Gao Z J, et al. 2017. A solar radiation prediction model based on radom forest. Control Engineering of China (in Chinese), 24(12): 2472-2477, doi: 10.14107/j.cnki.kzgc.150745.
Murata A, Ohtake H, Oozeki T. 2018. Modeling of uncertainty of solar irradiance forecasts on numerical weatherpredictions with the estimation of multiple confidence intervals. Renewable Energy, 117: 193-201, doi: 10.1016/j.renene.2017.10.043.
National Energy Administration. 2014. NB/T 320112013 Power Prediction System Technology of Photovoltaic Power Station (in Chinese). Beijing: Xinhua Publishing House.
Qing X, Niu Y. 2018. Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM. Energy, 148: 461-468, doi: 10.1016/j.energy.2018.01.177.
Sun H W, Gui D W, Yan B W, et al. 2016. Assessing the potential of random forest method for estimating solar radiation using air pollution index. Energy Conversion and Management, 119: 121-129, doi: 10.1016/j.enconman.2016.04.051.
Tashman L J. 2000. Out-of-sample tests of forecasting accuracy: an analysis and review. International Journal of Forecasting, 16(4): 437-450, doi: 10.1016/S0169-2070(00)00065-0.
Taylor S J, Letham B. 2018. Forecasting at scale. The American Statistician, 72(1): 37-45, doi: 10.1080/00031305.2017.1380080.
Thorey J, Mallet V, Chaussin C, et al. 2015. Ensemble forecast of solar radiation using TIGGE weather forecasts and HelioClim database. Solar Energy, 120: 232-243, doi: 10.1016/j.solener.2015.06.049.
Van Hoey S, van der Kwast J, Nopens I, et al. 2013. Python package for model structure analysis (pySTAN). Open Water, 2(1): 3.
Yi D H, Wang Y. 2019. Applied Time Series Analysis (in Chinese). 5th ed. Beijing: Renmin University of China Press.
Yue Y C, Huang T Z. 2007. A method for error reciprocal variable weight combined forecasting. Journal of University of Electronic Science and Technology of China (in Chinese), 36(2): 349-351.
Zang H X, Cheng L L, Liu L, et al. 2021. Research and prospect for data-driven estimation and prediction of solar radiation. Automation of Electric Power Systems (in Chinese), 45(11): 170-183, doi: 10.7500/AEPS20200709004.
Zeng Z L, Wang Z M, Gui K, et al. 2020. Daily global solar radiation in China estimated from high-density meteorological observations: A random forest model framework. Earth and Space Science, 7(2): e2019EA001058, doi: 10.1029/2019EA001058.
Zhang G P. 2003. Time series forecasting using a hybrid ARIMA and neural network model. NeuroComputing, 50: 159-175, doi: 10.1016/S0925-2312(01)00702-0.
Zhang J Y, Zhao L, Deng S, et al. 2017. A critical review of the models used to estimate solar radiation. Renewable and Sustainable Energy Reviews, 70: 314-329, doi: 10.1016/j.rser.2016.11.124.
Zhang M, Lu G P, Huang H L, et al. 2023. Automatic recognition of tweek atmospherics and plasma diagnostics in the lower ionosphere with the machine learning method. Earth and Planetary Physics, 7(3): 407-413, doi: 10.26464/epp2023039.
Zhang X S, Zhu X, Zhao B, et al. 2017. Short-term forecast of solar radiation based on WT-PLSP. Journal of Arid Land Resources and Environment (in Chinese), 31(6): 142-146, doi: 10.13448/j.cnki.jalre.2017.192.
Zhang Y, Chen Z. 2018. Study of improved ARIMA model in medical demand forecasting. Computer Engineering and Design (in Chinese), 39(1): 183-188, 295, doi: 10.16208/j.issn1000-7024.2018.01.032.
Zhang Z G, Qin X, He L F, et al. 2009. Radiation character of Mt. Qomolangma area during late spring to early summer in 2007. Plateau Meteorology (in Chinese), 28(3): 516-522.
国家能源局. 2014. NB/T 32011-2013光伏发电站功率预测系统技术要求. 北京: 新华出版社.
郝玉珠, 李兴华, 胡亚男等. 2021. 内蒙古57年太阳能资源变化规律及影响因子. 太阳能学报, 42(9): 145-151, doi: 10.19912/j.0254-0096.tynxb.2019-0868.
侯宁, 张晓通, 魏瑜等. 2021. 基于随机森林方法的中国地表短波辐射估算. 太阳能学报, 42(2): 31-36, doi: 10.19912/j.0254-0096.tynxb.2018-0989.
刘剑, 曹美燕, 高治军等. 2017. 一种基于随机森林的太阳能辐射预测模型. 控制工程, 24(12): 2472-2477, doi: 10.14107/j.cnki.kzgc.150745.
易丹辉, 王燕. 2019. 应用时间序列分析. 5版. 北京: 中国人民大学出版社.
岳艳春, 黄廷祝. 2007. 误差倒数变权组合预测方法. 电子科技大学学报, 36(2): 349-351. https://www.cnki.com.cn/Article/CJFDTOTAL-DKDX2007S1010.htm
臧海祥, 程礼临, 刘玲等. 2021. 基于数据驱动的太阳辐射估计和预测研究与展望. 电力系统自动化, 45(11): 170-183, doi: 10.7500/AEPS20200709004.
张雪松, 朱想, 赵波等. 2017. 基于WT-PLSP的太阳辐射短期预测. 干旱区资源与环境, 31(6): 142-146, doi: 10.13448/j.cnki.jalre.2017.192.
张颖, 陈卓. 2018. 改进ARIMA模型在医药需求预测中的研究. 计算机工程与设计, 39(1): 183-188, 295, doi: 10.16208/j.issn1000-7024.2018.01.032.
张志刚, 秦翔, 何立富等. 2009. 2007年春末夏初珠穆朗玛峰地区辐射特征分析. 高原气象, 28(3): 516-522. https://www.cnki.com.cn/Article/CJFDTOTAL-GYQX200903005.htm
-
图(8)
表(8)
计量
- 文章访问数: 1834
- PDF下载数: 120
- 施引文献: 0