Assessment of solar radiation resource and photovoltaic power potential across China based on optimized interpretable machine learning model and GIS-based approaches

Research output: Journal article publicationJournal articleAcademic researchpeer-review

21 Citations (Scopus)

Abstract

In light of the rapidly expanding solar photovoltaic (PV) sector, it is important to provide a deeper understanding of solar energy resources to successfully implement solar energy projects. In this study, an interpretable machine learning model based on extreme gradient boosting (XGBoost) optimized by particle swarm optimization (PSO) algorithms was developed to estimate global solar radiation. The results show that the proposed PSO-XGBoost model possesses the most superior accuracy and stability, with the coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) of 0.953, 1.597 MJ·m−2·day−1, 1.138 MJ·m−2·day−1, and 10.500%, respectively. With the geographic information system (GIS) -based approaches, a 50 km by 50 km spatial resolution map of long-term national average solar radiation resources was generated based on the reconstructed solar radiation dataset, as well as the PV power potential map. The findings reveal that the nationwide annual mean solar radiation resources were decreasing at an estimated attenuation of −0.83 W·m−2·decade−1, with a downward trend of the greatest magnitude of −1.83 W·m−2·decade−1 for summer. China's long-term average yearly PV power potential reached 285.00 kWh·m−2, indicating a spatial pattern of higher potentials in the northwestern and northern provinces, while lower values in the southeastern provinces. Moreover, the PV power potential in China decreased by 1.69 kWh·m−2·decade−1 from 1961 to 2016, with an attenuation of above 5 kWh·m−2·decade−1 in heavily polluted regions. During the 2010s, 30 out of the 31 provinces experienced a reduction in the PV power potential between 0.25% and 10.27%, with an average national reduction of 2.88%, compared to the 1960s scenario. Also, policy recommendations for long-term PV project deployment were given regarding the regional mismatch between PV power potential and installed capacity in China.

Original languageEnglish
Article number121005
JournalApplied Energy
Volume339
DOIs
Publication statusPublished - 1 Jun 2023

Keywords

  • Geographic information system
  • Machine learning
  • Solar photovoltaic power potential
  • Solar radiation resource
  • Spatiotemporal pattern

ASJC Scopus subject areas

  • Building and Construction
  • Mechanical Engineering
  • General Energy
  • Management, Monitoring, Policy and Law

Fingerprint

Dive into the research topics of 'Assessment of solar radiation resource and photovoltaic power potential across China based on optimized interpretable machine learning model and GIS-based approaches'. Together they form a unique fingerprint.

Cite this