Abstract:

A method of using traffic flow parameters is proposed for traffic status identification based on Monte Carlo simulation. Clustering analysis of four kinds traffic status is realized by applying FANNY algo- rithm, and SVC traffic status identification model is established using Monte Carlo simulation method. Fixed window model and sliding window model are built respectively to identify and conduct comprehensive evalu- ation on traffic status. Results indicate that the method can achieve accurate identification of real-time traffic flow parameters, especially with sliding window model, of which average identification accuracy, recall and F-measure are 97.98%, 94.64% and 96.21% respectively. It provides scientific methods and data support for analyzing evolution regularity and development trend of traffic status, as well as establishing emergency op- eration mechanism such as prediction and forewarning, emergency disposal and information release.

Key words: highway transportation, traffic status identification, Monte Carlo simulation, traffic flow pa- rameters, SVC, data mining

摘要：

提出了一种基于蒙特卡罗模拟的利用交通流参数实现交通状态辨识的方法.采 用 FANNY 算法实现了四种交通状态的聚类分析；利用蒙特卡罗模拟方法建立了 SVC 交 通状态辨识模型；分别构建了固定窗口模型和滑动窗口模型对交通状态进行辨识并综合 评价.分析结果表明：该方法能够对实时交通流参数进行准确辨识，尤其是构建的滑动窗 口模型，对交通状态辨识平均精度、召回率和 F 度量分别为 97.98%、94.64%和 96.21%.本方 法可为分析高速公路交通状态演化规律和发展趋势，建立预测预警、应急处置和信息发 布等应急运行机制提供科学方法和数据支撑.

关键词: 公路运输, 交通状态辨识, 蒙特卡罗模拟, 交通流参数, SVC, 数据挖掘