大数据中高速公路旅行时间预测仿真研究

摘要

旅行时间预测是智能交通研究中的重要内容,能够形象的反映道路的交通状况.准确的旅行时间预测可以为高速公路管理部门和出行者提供决策支持.而旅行时间预测需要可靠、完善的数据采集系统作为支撑,传统的数据采集方法不但难度大成本高,而且采集的数据量较少,准确度低.而高速公路收费数据准确记录了车辆进出的时间地点等信息,且有足够的数据量.高速公路的旅行时间具有非线性和不确定性的特征,传统的预测模型的预测结果准确性较低.根据旅行时间的这些特征,提出了一种小波神经网络的旅行时间预测模型.仿真结果显示在收费系统的大数据支持下,采用小波神经网络算法的预测结果比传统方法精度有了进一步的提高.%Travel time is an important factor of intelligent transportation system,which can reflect the traffic of highway.Accurate prediction of travel time can provide decision support for the highway management department and travelers.It needs the reliable and perfect information collection system as a support,and needs a lot of inputs.Traditional methods of data collection have many disadvantages,such as large difficulty,high cost,little data and low accuracy.Highway network toll data can record the time and location of every car,and the amount of data is enough.Highway travel time is characterized by high nonlinearity and uncertainty.The conventional prediction model has low accurate.This paper proposes a prediction model of travel time based on wavelet neural network.The travel time is predicted based on toll data and wavelet neural network.The method based on big data and wavelet neural network has a more accurate prediction result than traditional methods.