Predicting the Change of CO2 Emissions Using a BNN-FA Method: A Case Study of Hebei Province

An integrated method that can simulate the temporal trends of CO2 emissions is crucial for aligning with China’s dual carbon targets and achieving carbon reduction within Hebei Province. In this study, a BNN-FA method integrating Bayesian neural network and factorial analysis is developed to explore the temporal variation of CO2 emissions. BNN-FA has the advantages of quantifying the effects of various factors and their contributions and predicting future trends of dependent variables. Then, the BNN-FA method is applied to analyzing Hebei Province's CO2 emissions (HBCE) under 64 scenarios. The results indicate that: (i) the two factors that contribute the most are the consumption of fossil energy (COFE, 72.7%) and the consumption of non-fossil energy (CONF, 21.3%), representing the significant impact of the energy consumption on HBCE; (ii) the range of CO2 emissions reduction potential is 2.5 × 108 tonnes ~ 8.5 × 108 tonnes (t) under all scenarios, and HBCE shows downward trends during 2040–2060 when more negative high-level factors are involved (e.g., scenarios 43–55); (iii) under the optimal development mode (scenario 43), the maximum emissions reduction potential is 8.5 × 108 t and the industrial transformation rate is 253.1% compared to the extreme development mode (scenario 64).