Analysis and prediction of green hydrogen production potential by photovoltaic-powered water electrolysis using machine learning in China

With the background of energy saving and emission reduction, hydrogen production needs to be gradually changed from gray hydrogen to green hydrogen. Photovoltaic water electrolysis is a good method for producing green hydrogen, and its potential in China needs to be explored. The focus of this study is to analyze and predict the potential of green hydrogen production by photovoltaic-powered water electrolysis using machine learning methods in China. For this purpose, a photovoltaic-electrolysis (PV-E) system was selected to forecast the hydrogen production. Among the two selected algorithms, the non-time series algorithm SVM performs better than FbProphet. The R² values of the test sets in four regions are 0.968, 0.980, 0.955, and 0.960. The RMSE values of the test sets in four regions are 81.3 kg/km², 31.4 kg/km², 71.1 kg/km², and 70.0 kg/km². The predicted daily hydrogen production in the two regions with higher radiation, MPZ and TCZ, are 32.5 × 10³ kg/km² and 27.5 × 10³ kg/km², respectively. Combined with the unused land area in the two districts, it is expected that the annual hydrogen production can be as high as 8.3368 × 10⁹ t and 1.0259 × 10¹⁰ t. This study demonstrates the viability and environmental friendliness of green hydrogen generation using photovoltaic-powered water electrolysis in China.