我国“十四五”发展规划对交通绿色发展提出了更高要求，交通运营所产生的排放是交通行业碳排放的主要来源。为研究公路平曲线路段小客车运行碳排放的影响因素，通过开展现场实车试验，试验车载OBD（On Board Diagnostics）设备采集广东省内典型公路曲线段现场行车试验数据，并通过IPCC（Intergovernmental Panel on Climate Change）碳排放核算方法获取路段碳排放率数据，结合道路线形选取了影响小客车运行排放的相关评价指标，在灰色关联分析的基础上对相关评价指标进行关联度计算。结果表明：平曲线路段线形要素中缓和曲线长度占比、缓和曲线参数等指标与路段碳排放率显著关联；圆曲线半径在一定区间内与路段碳排放率显著关联。非线形指标中显著相关的有加速度标准差和加速度均值，与这两者显著相关的线形指标是缓和曲线参数和缓和曲线长度占比。结合灰色关联分析结果，从指标中选取了8个关联指标，通过建立灰色GM（1，N）模型对平曲线路段小客车行驶碳排放总量进行预测，预测结果和实际结果平均相对误差为5.10%，模型预测性能优于传统多元回归模型，在数据有限的情况下表现出色并提供可靠的预测结果。研究成果可识别路段对碳排放有显著影响的关键设计和运行参数，为平曲线路段低碳优化设计和管理提供理论依据。

China’s “14th Five-Year Plan” places higher demands on green transportation development, with emissions from traffic operations being the primary source of carbon emissions in the transportation sector. To investigate the factors influencing carbon emissions of passenger cars on highway curved segments, this study conducted on-site driving tests using OBD-equipped vehicles to collect driving data from typical curved road segments in Guangdong Province, and obtains carbon emission data through the IPCC carbon emission accounting method. Re-levant evaluation indicators influencing passenger car emissions were selected based on road alignment, and gray relational analysis was used to calculate the correlations between these indicators. The results indicate that among the geometric alignment elements of horizontal curve sections, indicators such as the proportion of transition curve length and transition curve parameters are significantly correlated with the segmental carbon emission rate. The radius of the circular curve is also significantly correlated within a specific range. For non-geometric factors, indicators such as the standard deviation and mean of acceleration show significant correlations with carbon emissions, and these two indicators are further associated with geometric factors like transition curve parameters and the proportion of transition curve length. Based on the results of the grey relational analysis, eight correlated indicators were selected, and a grey GM(1, N) model was developed to predict the total carbon emissions of passenger cars on horizontal curve sections. The prediction results show an average relative error of 5.10% compared to the actual values. The predictive performance of the model surpasses that of traditional multiple regression models, demonstrating outstanding performance and reliability in scenarios with limited data.The findings of this study can identify key design and operational parameters significantly influencing carbon emissions, providing a theoretical basis for the low-carbon optimization and management of highway horizontal curve sections.