基于NGSIM的车辆速度运行状态特征研究

Research on Vehicle Speed Operation State Characteristics Based on NGSIM

摘要: 基于NGSIM (Next Generation Simulation)项目US-101路段的微观轨迹数据，综合利用Python、SPSS及MATLAB软件，从宏观与微观层面深入分析了城市快速路车辆运行速度的分布规律及变化机理。统计分析表明，研究路段总体及各分车道的车速分布均服从正态分布，且低速车辆车速较低、高速车辆速度差异小，整体离散程度较弱。通过构建多项式曲线模型对不同车头间距与车道位置的速度特性进行拟合，发现三次多项式模型能较好地描述5~50 m跟驰状态下的速度–间距关系，其中第4车道拟合优度最高(R² = 0.995)；当车头间距小于5 m时，速度集中于低速区，而大于50 m时则进入自由流状态。车道差异分析揭示，小间距时最内侧车道(Lane 1)速度最低，随着间距增大至35 m以上，各车道速度趋于一致(约45 km/h)。此外，时空分布特征显示，受匝道与交织区影响，路段前300 m内的车速显著降低(最低至27.5 km/h)，通过辅道后速度明显跃升。该研究明确了正态分布对拥堵过渡状态下城市快速路车速的描述能力，确定了多项式模型描述跟驰行为的最佳间距范围，揭示了匝道与交织区对运行车速的显著影响机制，为交通流微观仿真与道路安全设计提供了理论依据与数据支撑。

Abstract: The distribution law and variation mechanism of vehicle operating speeds on urban expressways were analyzed at macroscopic and microscopic levels based on the microscopic trajectory data from the US-101 section of the Next Generation Simulation (NGSIM) project. Python, SPSS, and MATLAB software were comprehensively utilized for this analysis. Statistical results indicate that the speed distributions for the overall road segment and individual lanes conform to a normal distribution. Low-speed vehicles exhibit lower velocities, while high-speed vehicles show minimal speed variations, resulting in weak overall dispersion. Polynomial curve models were constructed to fit the speed characteristics under different headways and lane positions. The cubic polynomial model was found to describe the speed-headway relationship well during car-following states with headways between 5 m and 50 m. The fourth lane demonstrated the highest goodness of fit (R² = 0.995). When the headway is less than 5 m, speeds concentrate in the low-speed zone. Conversely, when the headway exceeds 50 m, traffic enters a free-flow state. Lane difference analysis reveals that the innermost lane (Lane 1) has the lowest speed at small headways. As the headway increases beyond 35 m, speeds across all lanes tend to converge (approximately 45 km/h). Furthermore, spatiotemporal distribution characteristics show that speeds within the first 300 m of the segment decrease significantly (dropping as low as 27.5 km/h) due to the influence of ramps and weaving areas. A significant speed surge occurs after passing the auxiliary road. This study clarifies the descriptive capability of the normal distribution for urban expressway speeds under congestion transition conditions. The optimal headway range for describing car-following behavior using the polynomial model is determined. The significant influence mechanism of ramps and weaving areas on operating speeds is revealed. These findings provide a theoretical basis and data support for traffic flow microsimulation and road safety design.