DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

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

Journal Article · · Solar Energy
 [1]; ORCiD logo [2];  [2];  [3];  [2];  [4];  [2]
  1. Brookhaven National Lab. (BNL), Upton, NY (United States); Nanjing University of Information Science and Technology (China); Hangzhou Meteorological Bureau (China)
  2. Brookhaven National Lab. (BNL), Upton, NY (United States)
  3. Nanjing University of Information Science and Technology (China)
  4. 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.

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; 33504; AC36-08GO28308
OSTI ID:
1887823
Report Number(s):
BNL-223344-2022-JAAM
Journal Information:
Solar Energy, Vol. 244; ISSN 0038-092X
Publisher:
ElsevierCopyright Statement
Country of Publication:
United States
Language:
English

References (48)

A new look at the statistical model identification journal December 1974
Daily Solar Energy Estimation for Minimizing Energy Storage Requirements in PV Power Plants journal April 2013
Machine Learning Interpretability: A Survey on Methods and Metrics journal July 2019
Climate Change: An Appraisal of Atmospheric Feedback Mechanisms Employing Zonal Climatology journal October 1976
Short term forecasting of solar radiation and power output of 89.6kWp solar PV power plant journal January 2021
Review of solar irradiance forecasting methods and a proposition for small-scale insular grids journal November 2013
Distribution of the Estimators for Autoregressive Time Series with a Unit Root journal June 1979
Data-driven scenario generation of renewable energy production based on controllable generative adversarial networks with interpretability 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 journal May 2018
The role of renewable energy in the global energy transformation journal April 2019
A new airmass independent formulation for the Linke turbidity coefficient journal September 2002
Solar forecasting methods for renewable energy integration journal December 2013
Improving the cloud initialization in WRF-Solar with enhanced short-range forecasting functionality: The MAD-WRF model journal June 2022
Extreme gradient boosting and deep neural network based ensemble learning approach to forecast hourly solar irradiance journal January 2021
The Mythos of Model Interpretability: In machine learning, the concept of interpretability is both important and slippery. journal June 2018
The interrelationship and characteristic distribution of direct, diffuse and total solar radiation journal July 1960
Use of physics to improve solar forecast: Physics-informed persistence models for simultaneously forecasting GHI, DNI, and DHI journal February 2021
Relationship between cloud radiative forcing, cloud fraction and cloud albedo, and new surface-based approach for determining cloud albedo journal January 2011
Identification of clear skies from broadband pyranometer measurements and calculation of downwelling shortwave cloud effects journal June 2000
Estimation of fractional sky cover from broadband shortwave radiometer measurements journal January 2006
Deep learning based forecasting of photovoltaic power generation by incorporating domain knowledge journal June 2021
Forecasting of global and direct solar irradiance using stochastic learning methods, ground experiments and the NWS database journal May 2011
Making the Black Box More Transparent: Understanding the Physical Implications of Machine Learning journal November 2019
Two-Stream Approximations to Radiative Transfer in Planetary Atmospheres: A Unified Description of Existing Methods and a New Improvement journal March 1980
Retrievals of thin cloud optical depth from a multifilter rotating shadowband radiometer journal January 2004
Cloud properties derived from surface MFRSR measurements and comparison with GOES results at the ARM SGP Site journal June 1996
Mapping daily global solar irradiation over Spain: A comparative study of selected approaches journal September 2011
Predicting the Future — Big Data, Machine Learning, and Clinical Medicine journal September 2016
Validation of short and medium term operational solar radiation forecasts in the US journal December 2010
Hourly day-ahead solar irradiance prediction using weather forecasts by LSTM journal April 2018
Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations journal February 2019
Solar position algorithm for solar radiation applications journal January 2004
Forecasting ground-level irradiance over short horizons: Time series, meteorological, and time-varying parameter models journal November 2017
A short-term solar radiation forecasting system for the Iberian Peninsula. Part 1: Models description and performance assessment journal January 2020
Anisotropic nonconservative scattering and the clouds of Venus journal January 1967
Climate change and the permafrost carbon feedback journal April 2015
A feature learning approach based on XGBoost for driving assessment and risk prediction journal August 2019
Irreversible climate change due to carbon dioxide emissions journal January 2009
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
Physically Interpretable Neural Networks for the Geosciences: Applications to Earth System Variability journal August 2020
Machine learning methods for solar radiation forecasting: A review journal May 2017
Comparing support vector regression for PV power forecasting to a physical modeling approach using measurement, numerical weather prediction, and cloud motion data journal October 2016
Observationally based evaluation of NWP reanalyses in modeling cloud properties over the Southern Great Plains journal June 2012
A new approach for simultaneously retrieving cloud albedo and cloud fraction from surface-based shortwave radiation measurements journal October 2013
Retrievals of cloud fraction and cloud albedo from surface-based shortwave radiation measurements: A comparison of 16 year measurements: Cloud fraction and cloud albedo journal July 2014
Hourly solar irradiance time series forecasting using cloud cover index journal December 2012
Verification of deterministic solar forecasts journal November 2020
Short-Term Load Forecasting Using EMD-LSTM Neural Networks with a Xgboost Algorithm for Feature Importance Evaluation journal August 2017