← 研踪 PaperTrailarXiv:2611.24853 · v1
Beyond corridor averages: Lane-level validation of microscopic freeway simulation with empirically parameterized arrivals
超越走廊平均值:基于经验参数化到达的高速公路微观仿真的车道级验证
首次提交 2026-11-19 02:13
版本更新 2026-11-19 02:13
雷达观测 2026-11-19 15:04
摘要
Traffic flow modeling is an essential part of civil planning around the world, and traffic simulation is an important component of the analyses that city planners must perform to ensure safe and efficient road networks for their citizens. This work advances microscopic traffic simulation through lane-level validation of speed and flow dynamics, addressing a critical gap in conventional macro-aggregated approaches. We systematically evaluate the trade-offs between numerical integration methods and vehicle arrival processes when reproducing empirical trajectory data from California's freeway network. Results demonstrate that numerical-integrator choice has negligible impact on simulation accuracy, with less than one percent variation across eight methods, while vehicle arrival-process modeling substantially affects fidelity. The shifted Erlang-2 distribution reduces flow-prediction error by approximately 28% compared to Poisson processes by enforcing a realistic minimum headway. Lane-level validation across five detector stations reveals lane-specific dynamics that aggregated metrics often obscure.
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-19 02:13 · 最新观测版本
Beyond corridor averages: Lane-level validation of microscopic freeway simulation with empirically parameterized arrivals
Traffic flow modeling is an essential part of civil planning around the world, and traffic simulation is an important component of the analyses that city planners must perform to ensure safe and efficient road networks for their citizens. This work advances microscopic traffic simulation through lane-level validation of speed and flow dynamics, addressing a critical gap in conventional macro-aggregated approaches. We systematically evaluate the trade-offs between numerical integration methods and vehicle arrival processes when reproducing empirical trajectory data from California's freeway network. Results demonstrate that numerical-integrator choice has negligible impact on simulation accuracy, with less than one percent variation across eight methods, while vehicle arrival-process modeling substantially affects fidelity. The shifted Erlang-2 distribution reduces flow-prediction error by approximately 28% compared to Poisson processes by enforcing a realistic minimum headway. Lane-level validation across five detector stations reveals lane-specific dynamics that aggregated metrics often obscure.
查看此版本解读 →查看此版本原文 ↗