基于WiFi探针数据的短时交通状态预测  

Short-term traffic state prediction based on WiFi probe data

作　　者：吴启用 廖嘉欣 兰小机[1] WU Qiyong;LIAO Jiaxin;LAN Xiaoji(School of Civil and Surveying&Mapping Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,Jiangxi,China)

机构地区：[1]江西理工大学土木与测绘工程学院,江西赣州341000

出　　处：《江西理工大学学报》2021年第4期11-18,共8页Journal of Jiangxi University of Science and Technology

基　　金：国家自然科学基金资助项目(41561085);江西省研究生创新基金资助项目(YC2019-S309)。

摘　　要：针对传统交通数据获取成本高,以及单参数输入的LSTM模型预测精度不高等问题,提出一种基于WiFi探针数据的短时交通状态预测方法。首先利用WiFi探针数据构建交通状态指数数据集,然后采用LSTM网络构建预测模型,并分析不同交通参数组合对模型预测精度的影响,最后比较不同模型对同一路段的预测性能。试验结果表明:交通流量会影响交通状态指数的预测精度,相比于仅考虑交通状态指数的LSTM网络,同时引入交通流量和上下游信息的LSTM网络的预测精度提升明显,其RMSE、MAE分别降低了11.89%,12.22%,R2提高了3.6%;另外,LSTM模型较SVR和GBDT模型具有更高的预测精度,证明了该方法的有效性。Considering the high cost of traditional traffic data acquisition and lower prediction accuracy of single-parameter input LSTM model,this paper has proposed a short-term traffic state prediction method based on WiFi probe data.Firstly,the traffic state index data set was constructed by using WiFi probe data,then the prediction model was constructed by LSTM network,and the influence of different traffic parameter combinations on the prediction accuracy of the model was analyzed.Finally,the prediction performances of different models for the same section were compared.The test results show that traffic flow can affect the prediction accuracy of traffic state index.Compared with the LSTM network which only considers traffic state index,the prediction accuracy of LSTM network which introduces traffic flow as well as upstream and downstream information is improved obviously,with its RMSE and MAE decreased by 11.89%and 12.22%respectively,and its R2 increased by 3.6%.In addition,LSTM model has higher prediction accuracy than SVR and GBDT model,which proves the effectiveness of this method.

关 键 词：交通拥堵 WiFi探针 LSTM网络 交通状态 

分 类 号：U495[交通运输工程—交通运输规划与管理]