Abstract: Using the transportation sector of Shandong Province as a case study, this research employed the Stochastic Impacts by Regression on Population, Affluence, and Technology (STIRPAT) model combined with ridge regression to analyze the driving factors of carbon emissions. The Particle Swarm Optimization (PSO) algorithm was used to optimize the Back Propagation (BP) neural network, constructing a combined PSO-BP neural network prediction model with seven variables, including population and per capita GDP, as the input layer. The model was applied to predict and analyze CO₂ emissions from the transportation sector of Shandong Province from 2023 to 2035 under three scenarios. The results indicate that population size, per capita GDP, energy structure, transportation energy intensity, freight turnover, and the number of civilian vehicles are promoting factors for carbon emissions in Shandong Province's transportation sector, while transportation intensity acts as a suppressing factor. The PSO-BP prediction model demonstrated high accuracy and a good fit. Compared with single BP neural networks, Support Vector Regression (SVR) models, and the STIRPAT model, the prediction results showed reductions in mean absolute percentage error by 5.78%, 2.00%, and 3.78%, root mean square error by 3.357×10⁶, 1.539×10⁶ and 1.953×10⁶ tons, and mean absolute error by 2.651×10⁶, 1.128×10⁶ and 1.756×10⁶ tons, respectively. During the forecast period, CO₂ emissions from Shandong Province's transportation sector are projected to peak at 5.535×10⁷ tons in 2030 under the low-carbon scenario, while maintaining an upward trend under the baseline and high-carbon scenarios. Based on existing policies, Shandong Province should continuously optimize its transportation structure, actively promote low-carbon travel methods, and increase the proportion of clean energy applications to achieve green, low-carbon, and high-quality development in the transportation sector.