基于深度学习和大数据分析的智慧交通流量预测模型研究  

Research on Intelligent Traffic Flow Prediction Model Based on Deep Learning and Big Data Analysis

作　　者：崔金魁[1] Cui Jinkui(School of Intelligent Transportation,Nanjing Vocational College of Information Technology,Nanjing 210023,China)

机构地区：[1]南京信息职业技术学院智能交通学院,南京210023

出　　处：《信息化研究》2024年第3期16-22,共7页INFORMATIZATION RESEARCH

基　　金：2023年江苏省高职院校“车路协同”智慧交通技术研究开发中心建设项目;2022年中国高校产学研创新基金项目(No.2022BC149)。

摘　　要：本文探讨了智慧交通系统中交通流量预测现状及深度学习在交通数据处理中的应用,在此基础上提出一种新的深度学习模型,专门用于智慧交通系统中的交通流量预测。该模型结合深度卷积神经网络和门控循环单元,可高效处理交通数据中的空间和时间特征。通过验证,该模型在不同城市、路段类型和天气条件下均展现出卓越的预测能力和强大的数据适应性。在主干道上,模型的准确率达到89.4%,均方误差(MSE)为0.045;在支路上,准确率达到82.7%~85.2%,MSE介于0.039~0.055之间。与传统预测方法相比,特别是在复杂交通场景下,本模型在准确率和计算效率方面均有显著提升。This paper explores the current situation of traffic flow prediction in smart traffic systems and the application of deep learning in traffic data processing.Based on this,a new deep learning model is proposed,specifically for traffic flow prediction in smart traffic systems.This model combines deep convolutional neural networks and gated recurrent units to efficiently process spatial and temporal features in traffic data.Through verification,this model demonstrates excellent predictive ability and strong data adaptability in different cities,road types,and weather conditions.On the main road,the accuracy of this model reaches 89.4%,with a mean square error(MSE)of 0.045.On the branch,the accuracy reaches 82.7%~85.2%,and the MSE ranges from 0.039 to 0.055.Compared with traditional prediction methods,especially in complex traffic scenarios,this model has significantly improved accuracy and computational efficiency.

关 键 词：智慧交通系统 交通流量预测 深度学习 卷积神经网络 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]