In the low-carbon transition of green supply chains, only continuous technological innovation can turn challenges into opportunities. However, significant costs and externalities associated with research and development may undermine companies’ willingness to invest in low-carbon R&D. This paper comprehensively considers transportation costs, transshipment costs, refrigeration costs, and carbon emissions factors to construct a green logistics carbon emissions prediction model based on the ARIMA model. It calculates the carbon emissions from railway freight transportation in the Beijing-Tianjin-Hebei region from 2004 to 2021. The results indicate that the ARIMA model has good predictive performance and can be used to predict future carbon emissions from green railway freight transportation in the Beijing-Tianjin-Hebei region. A carbon reduction path optimization model is established with the objective of minimizing the sum of fixed costs, transportation costs, refrigeration costs, carbon emissions costs, and time window penalty costs. By combining the NSGA-II algorithm with a neighborhood search algorithm, an improved NSGA-II algorithm model is designed. This algorithm is used to solve path scheduling schemes for carbon reduction path optimization model distribution and standalone path distribution under different customer scales. Experimental results show that compared to Path 1, Path 2, and Path 3, the carbon emission reduction path optimization model reduces emissions by 19.25%, 9.75%, and 7.80%, respectively. When the customer scale is large, using the carbon emission reduction path optimization model is more advantageous for cost savings.
