Use of physics to improve solar forecast: Part II, machine learning and model interpretability

Liu, Weijia; Liu, Yangang; Zhang, Tao; Han, Yongxiang; Zhou, Xin; Xie, Yu; Yoo, Shinjae

doi:10.1016/j.solener.2022.08.040

Use of physics to improve solar forecast: Part II, machine learning and model interpretability

Journal Article · Thu Sep 01 00:00:00 EDT 2022 · Solar Energy

DOI:https://doi.org/10.1016/j.solener.2022.08.040· OSTI ID:1887823

Liu, Weijia ^[1]; ^[2]; Zhang, Tao ^[2]; Han, Yongxiang ^[3]; Zhou, Xin ^[2]; Xie, Yu ^[4]; Yoo, Shinjae ^[2]

Brookhaven National Lab. (BNL), Upton, NY (United States); Nanjing University of Information Science and Technology (China); Hangzhou Meteorological Bureau (China)
Brookhaven National Lab. (BNL), Upton, NY (United States)
Nanjing University of Information Science and Technology (China)
National Renewable Energy Lab. (NREL), Golden, CO (United States)

We report machine learning (ML) models have been applied to forecast solar energy; however, they often lack clarity of interpretability and underlying physics. This work addresses such challenges by developing a hierarchy of ML models that gradually introduce predictors to improve the forecast accuracy based on a physics-based framework. Three ML models (ARIMA, LSTM, and XGBoost) are examined and compared with four physics-informed persistence models reported in Part I and the simple persistence model to assess the improvement of different models. The 7-year measurements at the U.S. Department of Energy's Atmospheric Radiation Measurement's Southern Great Plains Central Facility site are used for forecasts and evaluations. The results reveal that the step-by-step introduction of predictors leads to different improvements for models at different hierarchical levels. Comparison of the ML models with persistence models shows that LSTM and XGBoost outperform all the persistence models, with LSTM having the overall best performance; however, ARIMA underperforms the four physics-informed persistence models. This study demonstrates the importance and utility of incorporating physics into ML models in improving forecast accuracy by introducing a hierarchy of physics-based predictors, distinguishing predictor contributions, and enhancing the ML interpretability. The combined use of Global Horizontal Irradiance (GHI) and Direct Normal Irradiance (DNI) significantly improves the forecast accuracy compared to using individual irradiances alone because the pair contains more information on cloud-radiation interactions.

View Accepted Manuscript (DOE)

Research Organization:: Brookhaven National Laboratory (BNL), Upton, NY (United States); National Renewable Energy Laboratory (NREL), Golden, CO (United States)

Sponsoring Organization:: USDOE Office of Energy Efficiency and Renewable Energy (EERE), Renewable Power Office. Solar Energy Technologies Office; USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)

Grant/Contract Number:: SC0012704; AC36-08GO28308

OSTI ID:: 1887823

Alternate ID(s):: OSTI ID: 1890143

Report Number(s):: BNL-223344-2022-JAAM

Journal Information:: Solar Energy, Journal Name: Solar Energy Vol. 244; ISSN 0038-092X

Publisher:: ElsevierCopyright Statement

Country of Publication:: United States

Language:: English

References (48)

Retrievals of cloud fraction and cloud albedo from surface-based shortwave radiation measurements: A comparison of 16 year measurements: Cloud fraction and cloud albedo Xie, Yu; Liu, Yangang; Long, Charles N. Journal of Geophysical Research: Atmospheres, Vol. 119, Issue 14 https://doi.org/10.1002/2014JD021705	journal	July 2014
The interrelationship and characteristic distribution of direct, diffuse and total solar radiation Liu, Benjamin Y. H.; Jordan, Richard C. Solar Energy, Vol. 4, Issue 3 https://doi.org/10.1016/0038-092X(60)90062-1	journal	July 1960
A new airmass independent formulation for the Linke turbidity coefficient Ineichen, Pierre; Perez, Richard Solar Energy, Vol. 73, Issue 3 https://doi.org/10.1016/S0038-092X(02)00045-2	journal	September 2002
A feature learning approach based on XGBoost for driving assessment and risk prediction Shi, Xiupeng; Wong, Yiik Diew; Li, Michael Zhi-Feng Accident Analysis & Prevention, Vol. 129 https://doi.org/10.1016/j.aap.2019.05.005	journal	August 2019
Data-driven scenario generation of renewable energy production based on controllable generative adversarial networks with interpretability Dong, Wei; Chen, Xianqing; Yang, Qiang Applied Energy, Vol. 308 https://doi.org/10.1016/j.apenergy.2021.118387	journal	February 2022
Comparison of Support Vector Machine and Extreme Gradient Boosting for predicting daily global solar radiation using temperature and precipitation in humid subtropical climates: A case study in China Fan, Junliang; Wang, Xiukang; Wu, Lifeng Energy Conversion and Management, Vol. 164 https://doi.org/10.1016/j.enconman.2018.02.087	journal	May 2018
Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM Qing, Xiangyun; Niu, Yugang Energy, Vol. 148 https://doi.org/10.1016/j.energy.2018.01.177	journal	April 2018
Deep learning based forecasting of photovoltaic power generation by incorporating domain knowledge Luo, Xing; Zhang, Dongxiao; Zhu, Xu Energy, Vol. 225 https://doi.org/10.1016/j.energy.2021.120240	journal	June 2021
The role of renewable energy in the global energy transformation Gielen, Dolf; Boshell, Francisco; Saygin, Deger Energy Strategy Reviews, Vol. 24 https://doi.org/10.1016/j.esr.2019.01.006	journal	April 2019
Extreme gradient boosting and deep neural network based ensemble learning approach to forecast hourly solar irradiance Kumari, Pratima; Toshniwal, Durga Journal of Cleaner Production, Vol. 279 https://doi.org/10.1016/j.jclepro.2020.123285	journal	January 2021
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations Raissi, M.; Perdikaris, P.; Karniadakis, G. E. Journal of Computational Physics, Vol. 378 https://doi.org/10.1016/j.jcp.2018.10.045	journal	February 2019
Short term forecasting of solar radiation and power output of 89.6kWp solar PV power plant Das, Subhra Materials Today: Proceedings, Vol. 39 https://doi.org/10.1016/j.matpr.2020.08.449	journal	January 2021
Solar forecasting methods for renewable energy integration Inman, Rich H.; Pedro, Hugo T. C.; Coimbra, Carlos F. M. Progress in Energy and Combustion Science, Vol. 39, Issue 6 https://doi.org/10.1016/j.pecs.2013.06.002	journal	December 2013
Machine learning methods for solar radiation forecasting: A review Voyant, Cyril; Notton, Gilles; Kalogirou, Soteris Renewable Energy, Vol. 105 https://doi.org/10.1016/j.renene.2016.12.095	journal	May 2017
Forecasting ground-level irradiance over short horizons: Time series, meteorological, and time-varying parameter models Reikard, Gordon; Haupt, Sue Ellen; Jensen, Tara Renewable Energy, Vol. 112 https://doi.org/10.1016/j.renene.2017.05.019	journal	November 2017
Review of solar irradiance forecasting methods and a proposition for small-scale insular grids Diagne, Maimouna; David, Mathieu; Lauret, Philippe Renewable and Sustainable Energy Reviews, Vol. 27 https://doi.org/10.1016/j.rser.2013.06.042	journal	November 2013
Solar position algorithm for solar radiation applications Reda, Ibrahim; Andreas, Afshin Solar Energy, Vol. 76, Issue 5 https://doi.org/10.1016/j.solener.2003.12.003	journal	January 2004
Validation of short and medium term operational solar radiation forecasts in the US Perez, Richard; Kivalov, Sergey; Schlemmer, James Solar Energy, Vol. 84, Issue 12 https://doi.org/10.1016/j.solener.2010.08.014	journal	December 2010
Forecasting of global and direct solar irradiance using stochastic learning methods, ground experiments and the NWS database Marquez, Ricardo; Coimbra, Carlos F. M. Solar Energy, Vol. 85, Issue 5 https://doi.org/10.1016/j.solener.2011.01.007	journal	May 2011
Mapping daily global solar irradiation over Spain: A comparative study of selected approaches Moreno, A.; Gilabert, M. A.; Martínez, B. Solar Energy, Vol. 85, Issue 9 https://doi.org/10.1016/j.solener.2011.05.017	journal	September 2011
Hourly solar irradiance time series forecasting using cloud cover index Yang, Dazhi; Jirutitijaroen, Panida; Walsh, Wilfred M. Solar Energy, Vol. 86, Issue 12 https://doi.org/10.1016/j.solener.2012.07.029	journal	December 2012
Comparing support vector regression for PV power forecasting to a physical modeling approach using measurement, numerical weather prediction, and cloud motion data Wolff, Björn; Kühnert, Jan; Lorenz, Elke Solar Energy, Vol. 135 https://doi.org/10.1016/j.solener.2016.05.051	journal	October 2016
A short-term solar radiation forecasting system for the Iberian Peninsula. Part 1: Models description and performance assessment Rodríguez-Benítez, Francisco J.; Arbizu-Barrena, Clara; Huertas-Tato, Javier Solar Energy, Vol. 195 https://doi.org/10.1016/j.solener.2019.11.028	journal	January 2020
Verification of deterministic solar forecasts Yang, Dazhi; Alessandrini, Stefano; Antonanzas, Javier Solar Energy, Vol. 210 https://doi.org/10.1016/j.solener.2020.04.019	journal	November 2020
Use of physics to improve solar forecast: Physics-informed persistence models for simultaneously forecasting GHI, DNI, and DHI Liu, Weijia; Liu, Yangang; Zhou, Xin Solar Energy, Vol. 215 https://doi.org/10.1016/j.solener.2020.12.045	journal	February 2021
Improving the cloud initialization in WRF-Solar with enhanced short-range forecasting functionality: The MAD-WRF model Jiménez, Pedro A.; Dudhia, Jimy; Thompson, Gregory Solar Energy, Vol. 239 https://doi.org/10.1016/j.solener.2022.04.055	journal	June 2022
Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects Long, Charles N.; Ackerman, Thomas P. Journal of Geophysical Research: Atmospheres, Vol. 105, Issue D12 https://doi.org/10.1029/2000JD900077	journal	June 2000
Retrievals of thin cloud optical depth from a multifilter rotating shadowband radiometer Min, Qilong Journal of Geophysical Research, Vol. 109, Issue D2 https://doi.org/10.1029/2003JD003964	journal	January 2004
Estimation of fractional sky cover from broadband shortwave radiometer measurements Long, C. N.; Ackerman, T. P.; Gaustad, K. L. Journal of Geophysical Research, Vol. 111, Issue D11 https://doi.org/10.1029/2005JD006475	journal	January 2006
Observationally based evaluation of NWP reanalyses in modeling cloud properties over the Southern Great Plains Wu, Wei; Liu, Yangang; Betts, Alan K. Journal of Geophysical Research: Atmospheres, Vol. 117, Issue D12 https://doi.org/10.1029/2011JD016971	journal	June 2012
Physically Interpretable Neural Networks for the Geosciences: Applications to Earth System Variability Toms, Benjamin A.; Barnes, Elizabeth A.; Ebert‐Uphoff, Imme Journal of Advances in Modeling Earth Systems, Vol. 12, Issue 9 https://doi.org/10.1029/2019MS002002	journal	August 2020
Cloud properties derived from surface MFRSR measurements and comparison with GOES results at the ARM SGP Site Min, Qilong; Harrison, Lee C. Geophysical Research Letters, Vol. 23, Issue 13 https://doi.org/10.1029/96GL01488	journal	June 1996
Anisotropic nonconservative scattering and the clouds of Venus Sagan, Carl; Pollack, James B. Journal of Geophysical Research, Vol. 72, Issue 2 https://doi.org/10.1029/JZ072i002p00469	journal	January 1967
Climate change and the permafrost carbon feedback Schuur, E. A. G.; McGuire, A. D.; Schädel, C. Nature, Vol. 520, Issue 7546 https://doi.org/10.1038/nature14338	journal	April 2015
Predicting the Future — Big Data, Machine Learning, and Clinical Medicine Obermeyer, Ziad; Emanuel, Ezekiel J. New England Journal of Medicine, Vol. 375, Issue 13 https://doi.org/10.1056/NEJMp1606181	journal	September 2016
Irreversible climate change due to carbon dioxide emissions Solomon, Susan; Plattner, Gian-Kasper; Knutti, Reto Proceedings of the National Academy of Sciences, Vol. 106, Issue 6 https://doi.org/10.1073/pnas.0812721106	journal	January 2009
Distribution of the Estimators for Autoregressive Time Series with a Unit Root Dickey, David A.; Fuller, Wayne A. Journal of the American Statistical Association, Vol. 74, Issue 366a https://doi.org/10.1080/01621459.1979.10482531	journal	June 1979
A new approach for simultaneously retrieving cloud albedo and cloud fraction from surface-based shortwave radiation measurements Xie, Yu; Liu, Yangang Environmental Research Letters, Vol. 8, Issue 4 https://doi.org/10.1088/1748-9326/8/4/044023	journal	October 2013
A new look at the statistical model identification Akaike, H. IEEE Transactions on Automatic Control, Vol. 19, Issue 6 https://doi.org/10.1109/TAC.1974.1100705	journal	December 1974
Daily Solar Energy Estimation for Minimizing Energy Storage Requirements in PV Power Plants Beltran, Hector; Perez, Emilio; Aparicio, Néstor IEEE Transactions on Sustainable Energy, Vol. 4, Issue 2 https://doi.org/10.1109/TSTE.2012.2206413	journal	April 2013
The Mythos of Model Interpretability: In machine learning, the concept of interpretability is both important and slippery. Lipton, Zachary C. Queue, Vol. 16, Issue 3 https://doi.org/10.1145/3236386.3241340	journal	June 2018
Climate Change: An Appraisal of Atmospheric Feedback Mechanisms Employing Zonal Climatology Cess, Robert D. Journal of the Atmospheric Sciences, Vol. 33, Issue 10 https://doi.org/10.1175/1520-0469(1976)033<1831:CCAAOA>2.0.CO;2	journal	October 1976
Two-Stream Approximations to Radiative Transfer in Planetary Atmospheres: A Unified Description of Existing Methods and a New Improvement Meador, W. E.; Weaver, W. R. Journal of the Atmospheric Sciences, Vol. 37, Issue 3 https://doi.org/10.1175/1520-0469(1980)037<0630:TSATRT>2.0.CO;2	journal	March 1980
Making the Black Box More Transparent: Understanding the Physical Implications of Machine Learning McGovern, Amy; Lagerquist, Ryan; John Gagne, David Bulletin of the American Meteorological Society, Vol. 100, Issue 11 https://doi.org/10.1175/BAMS-D-18-0195.1	journal	November 2019
Machine Learning Interpretability: A Survey on Methods and Metrics Carvalho, Diogo V.; Pereira, Eduardo M.; Cardoso, Jaime S. Electronics, Vol. 8, Issue 8 https://doi.org/10.3390/electronics8080832	journal	July 2019
Short-Term Load Forecasting Using EMD-LSTM Neural Networks with a Xgboost Algorithm for Feature Importance Evaluation Zheng, Huiting; Yuan, Jiabin; Chen, Long Energies, Vol. 10, Issue 8 https://doi.org/10.3390/en10081168	journal	August 2017
Time series forecasting on multivariate solar radiation data using deep learning (LSTM) Sorkun, Murat Cihan; Durmaz İNcel, Özlem; Paoli, Christophe TURKISH JOURNAL OF ELECTRICAL ENGINEERING & COMPUTER SCIENCES, Vol. 28, Issue 1 https://doi.org/10.3906/elk-1907-218	journal	January 2020
Relationship between cloud radiative forcing, cloud fraction and cloud albedo, and new surface-based approach for determining cloud albedo Liu, Y.; Wu, W.; Jensen, M. P. Atmospheric Chemistry and Physics, Vol. 11, Issue 14 https://doi.org/10.5194/acp-11-7155-2011	journal	January 2011

Similar Records

Use of Physics to Improve Solar Forecast: Part II, Machine Learning and Model Interpretability

Journal Article · Thu Sep 01 00:00:00 EDT 2022 · Solar Energy · OSTI ID:1890143

Use of Physics to Improve Solar Forecast: Physics-Informed Persistence Models for Simultaneously Forecasting GHI, DNI, and DHI

Journal Article · Fri Jan 15 23:00:00 EST 2021 · Solar Energy · OSTI ID:1755143

Related Subjects

54 ENVIRONMENTAL SCIENCES
interpretability
machine learning models
persistence models
predictor contribution
solar irradiances

Use of physics to improve solar forecast: Part II, machine learning and model interpretability

Citation Formats

References (48)

Similar Records

Related Subjects