动态小波变换网络的短时交通流量预测  

Short-term Traffic Flow Prediction Based on Dynamic Wavelet Transform Network

作　　者：王习昇 杨春亭[1] 王海江 Wang Xisheng;Yang Chunting;Wang Haijiang(Zhejiang University of Science and Technology,Hangzhou 310023,China)

机构地区：[1]浙江科技学院,杭州310023

出　　处：《单片机与嵌入式系统应用》2020年第11期32-35,39,共5页Microcontrollers & Embedded Systems

基　　金：浙江省自然科学基金——面向物联网的密文检索技术研究(LQ20F020010)。

摘　　要：本文提出了一种基于神经网络的流量预测方法——动态小波变换网络模型,该模型结合了动态小波变换和门控递归单元。动态小波变换模块负责学习交通流量中的复杂特征关系,门控递归单元则利用学习交通数据的动态变化来捕捉时间依赖关系。在基于城市道路的交通预测上进行应用,实验结果表明,DWNN模型能够有效获得交通数据的时间相关性,更好地反映城市交通流量的变化特征。In the paper,a traffic prediction method is proposed,which is based on neural network-dynamic wavelet transform network(DWNN)model,it combines dynamic wavelet transform and gated recursive unit(LSTM).Specifically,the dynamic wavelet transform module is responsible for learning the complex feature relations in traffic flow,while the gated recursive unit captures the time dependence relations by learning the dynamic changes of traffic data.The experiment results show that the DWNN model can obtain the temporal correlation of traffic data efficiently and effectively and better reflect the changing characteristics of urban traffic flow.

关 键 词：智能交通 流量预测 小波变换 门控递归单元 动态小波变换网络 

分 类 号：U41[交通运输工程—道路与铁道工程] TP393[自动化与计算机技术—计算机应用技术]