利用BP神经网络预估拥堵路段通行时间的研究

Research on Estimating Traffic Time of Congested Road Section by BP Neural Network

为了更准确地预测从出发到目的地的出行时间,选择一个有代表性的路段进行数据收集和分析。首先将交通流状态分为拥堵、阻塞和平滑三大类,然后利用神经网络理论建立交通流特征参数和出行时间模型。经过两种神经网络模型的运行对比,时间序列的神经网络模型能够更好地预测车辆。旅行时间的准确度可达85%。

In order to predict the travel time from departure to destination more accurately,a representative road section is selected for data collection and analysis.First,the traffic flow state is divided into three categories:congestion,stoppage and smoothing,and then the neural network theory is used to establish traffic flow characteristic parameters and travel time models.After comparing the operation of the two neural network models,the time-series neural network model can better predict vehicles.Accuracy of travel time can reach 85%.