基于多层复杂网络的循环神经网络交通量预测模型  

Recurrent neural network traffic volume prediction model based on multi-layer complex network

作　　者：温志勇 翁小雄[1] 谢帮权 WEN Zhiyong;WENG Xiaoxiong;XIE Bangquan(South China University of Technology,Guangzhou 510641,China)

机构地区：[1]华南理工大学,广东广州510641

出　　处：《现代电子技术》2024年第22期173-178,共6页Modern Electronics Technique

基　　金：国家自然科学基金项目(52072129);广东省重点领域研发计划项目(2022B0101070001)。

摘　　要：针对未安装车流量检测设备的高速公路路段进行短时交通量准确预测,是一个亟待解决的问题。为此,提出一种基于复杂网络的循环神经网络路段短时交通量预测模型。该模型以入口节点交通量为输入,输出路段动态预测交通量。模型由复杂网络、交通小区划分、循环神经网络三个模块组成。复杂网络由多层网络组成,是交通小区划分的基础;交通小区划分模块根据节点特征值,采用聚类方法将节点形成小区,使同小区内节点具有类似特征。最后,以交通小区为依据,将节点交通量合并为小区交通量,采用循环神经网络进行路段动态交通量的预测。通过模型示例并与其他模型预测结果进行对比分析,验证所提模型的准确性和可靠性。结果表明,该模型能够准确地预测不同时长的交通量,MAPE为9.275%,相比于其他方法,预测精度更高且性能稳定,具有重要的应用价值。Accurate short-term traffic volume prediction for highway segments without installed traffic flow detection equipment is a pressing issue.On this basis,a recurrent neural network short-term traffic volume prediction model based on complex network is proposed.In this model,the traffic volume of the entry node is taken as input,and the traffic volume of the output section as dynamic prediction.The model is composed of three modules:complex network,traffic area division and recurrent neural network.The complex network is composed of multi-layer network,which is the basis of traffic district division.According to the characteristic value of the nodes,the traffic plot division module can use clustering method to form the nodes into a plot,so that the nodes in the same plot have similar characteristics.Based on the traffic area,the nodal traffic volume is combined into the traffic area volume,and the cyclic neural network is used to predict the dynamic traffic volume of the road section.The accuracy and reliability of the proposed model are verified by the model example and comparison with other model prediction results.The results show that the model can accurately predict the traffic volume of different time lengths,and the average MAPE is 9.275%.In comparison with other methods,the prediction accuracy is higher and the performance is stable,which has important application value.

关 键 词：交通量预测 高速公路路段 多层复杂网络 循环神经网络 交通小区划分 预测精度 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP391.9[电子电信—信息与通信工程]