摘要
精确的交通流量预测是实现未来智能交通的关键技术。神经网络模型在该领域的预测方面具有一定的优势。因此,为了提高预测精度,设计一种基于深度卷积神经网络的交通流量预测数学模型。首先,对交通流量数据的预处理方法进行分析,然后结合特征训练过程和卷积神经网络构建深度神经网络结构,并给出深度神经网络的配置参数。利用美国明尼苏达大学UMD分校的交通流数据集进行仿真实验,结果表明,提出的模型可以对短时交通全局趋势进行预测,并具有较好的稳定性和预测精度。
Accurate traffic flow prediction is a key technology for realizing intelligent transportation in the future. Neural network models have certain advantages in the prediction of this field. Therefore,in order to improve the prediction accuracy,a traffic flow prediction mathematical model based on deep convolutional neural network is designed. The preprocessing method of traffic flow data is analyzed. The deep neural network structure is constructed by combining feature training process and convolutional neural network. The configuration parameters of deep neural network are given in this paper. The simulation experiments were carried out with the traffic flow dataset of the UMD Branch,University of Minnesota. The results show that the proposed model can predict the short-term global traffic trend,and has good stability and prediction accuracy.
作者
刘红敏
LIU Hongmin(Guangzhou University Sontan College,Guangzhou 511370,China)
出处
《现代电子技术》
北大核心
2019年第13期110-112,共3页
Modern Electronics Technique
关键词
交通流量预测
智能交通
数学模型
深度神经网络
预测精度
仿真实验
traffic flow prediction
intelligent transportation
prediction methematical model
deep neural network
prediction accuracy
simulation experiment
