不同影响因素下的城市交通流预测  

Urban Traffic Flow Prediction Considering Different Factors

作　　者：李雪梅 梁建娟 刘洪 韩云杰 刘本永 LI Xuemei;LIANG Jianjuan;LIU Hong;HAN Yunjie;LIU Benyong(College of Big Data and Information Engineering,Guizhou University,Guiyang Guizhou 550025;Guiyang Institute of Information Technology,Guiyang Guizhou 550000)

机构地区：[1]贵州大学大数据与信息工程学院,贵州贵阳550025 [2]贵阳信息技术研究院,贵州贵阳550000

出　　处：《软件》2020年第12期67-71,共5页Software

基　　金：贵州省基金(黔科合基础[2019]1063号);贵州大学引进人才科研项目(贵大人基合同字(2017)13号、14号);国家自然科学基金项目(60862003)。

摘　　要：城市交通流具有非线性、相关性、周期性等特点,其建模和准确预测难度较大。为提高预测准确率,在雨天、非雨天和工作日、非工作日四种影响因素下搭建基于长短期记忆(LSTM)网络模型的双向长短期记忆(BiLSTM)网络模型用于提取不同因素下的数据周期特性,利用实测数据进行交通流预测分析实验。结果表明,与不考虑影响因素即随机环境相比,在考虑影响因素下BiLSTM模型预测的效果稍好;同时,在四个影响因素下,与传统RNN模型、LSTM模型进行比较,BiLSTM模型预测效果均稍好。Urban traffic flow has the characteristics of non-linearity,correlation and periodicity,so it is difficult to build model and predict urban traffic flow accurately.In order to improve the prediction accuracy,a Bidirectional Long Short-Term Memory(BiLSTM)network model based on Long Short-Term Memory(LSTM)network model is built to extract the data cycle characteristics in different factors,and the traffic flow prediction and analysis experiment was carried out using the measured data.Experimental results show that the prediction effect of BiLSTM model is slightly better than that of random environment without considering the influence factors.At the same time,Compared with the traditional RNN model and LSTM model,the prediction effect of BiLSTM model is slightly better under the four influence factors.

关 键 词：交通流预测 LSTM模型 BiLSTM模型 RNN模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]