基于车联网的实时路况估计方法

A Real-time Road Condition Estimation Method Based on Internet of Vehicles

针对实时路况估计和交通数据采集方法没有统一标准的问题,提出了一种基于车联网的实时路况估计架构,由车辆数据采集、数据管理和路况显示组成。车辆数据采集系统基于Android系统开发,采用模块化架构使系统适用于不同厂商的车型;通过对武汉市部分道路进行VISSIM建模,采集大量仿真数据,分别建立RBF神经网络模型和支持向量机模型,并使用遗传算法对支持向量机参数进行优化,两种模型估计结果表明支持向量机的估计效果优于RBF神经网络。最后将支持向量机估计结果应用于路况显示系统,向社会公众提供实时路况。

In this paper, in view of the lack of uniform standard in real-time road condition estimation and traffic data collection, we proposed a real- time road condition estimation framework based on the Internet of Vehicles which was composed by the components of vehicle data collection, data management and road condition readout. The vehicle data collection component was developed based on the Android platform, whose modular architecture rendered it suitable to accommodate the models of different makes. Then through the VISSIM modeling of several highways of Wuhan, we respectively built the RBF neural network model and SVM model of the city and used the genetic algorithm to optimize the parameterization of the support vector machine. At the end, through a numerical example, we demonstrated the superiority of the support vector machine over the RBF neural network in this application.