基于MLP的相关路段流量预测模型  

Traffic Flow Forecasting Model of Correlated Roads Based on MLP

作　　者：张弛[1] 付相君 周先颖 陈坚[1] ZHANG Chi;FU Xiangjun;ZHOU Xianying;CHEN Jian(School of Traffic and Transportation,Chongqing Jiaotong University,Chongqing 400074,China;Chongqing Transport Planning and Technology Development Center,Chongqing 400060,China)

机构地区：[1]重庆交通大学交通运输学院,重庆400074 [2]重庆市交通规划和技术发展中心,重庆400060

出　　处：《重庆理工大学学报（自然科学）》2021年第8期129-135,共7页Journal of Chongqing University of Technology：Natural Science

基　　金：重庆市社会科学规划重点项目(2020ZDZX04);重庆市教育委员会科学技术研究计划项目(KJQN202001611)。

摘　　要：为解决城市路网流量检测器无法达到路网全覆盖,部分道路无法得到有效监控成为交通数据采集系统“盲区”的问题,通过多维标度法(MDS)分析了城市路网中的路段流量相关性,并根据流量相对相关程度进行相关路段划分,针对每组相关路段基于多层感知机(MLP)建立结构为(6,p,q,1)的双隐层流量实时预测模型及(6,p,1)的单隐层对比模型。最后,将模型运用于重庆实例分析中,采用BP算法进行训练。结果表明:MLP模型实现了由已知路段流量至同组其他所有未知相关路段流量的整合建模,建模过程较传统实时流量预测方法更为简洁;其中双隐层模型预测精度较单隐层模型预测精度更高,达到87.4%,可指导路网流量检测器的布设优化。In order to solve the problem that some roads become the‘blind area’of traffic data acquisition system because the traffic detector of urban road network cannot reach the full coverage of road network,the correlation of traffic flow in urban road network is analyzed by multi-dimensional scaling(MDS)method,and the correlated road sections are divided according to the relative correlation degree of traffic flow.For each group of correlated road sections,the double hidden layer traffic flow forecasting model based on the multi-layer perceptron(MLP)with the structure of(6,p,q,1)and the single hidden layer comparison model with the structure of(6,p,1)are established.Finally,the models are applied to Chongqing case analysis and BP algorithm is used for training.The results show that:the MLP model realizes the integrated modeling of the traffic flow from the known road to all other unknown correlated roads in the same section,and the modeling process is more concise than the traditional real-time traffic forecasting methods.In addition,the forecasting accuracy of double hidden layer model is higher than that of single hidden layer model,reaching 87.4%,which can guide the layout optimization of road network flow detector.

关 键 词：智能交通系统 实时流量预测 多层感知机 流量相关性 

分 类 号：U491.14[交通运输工程—交通运输规划与管理]