[1]陈 刚.基于改进 GRU 的高速公路交通流量预测模型[J].计算机技术与发展,2023,33(07):208-214.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 031]
　CHEN Gang.Highway Traffic Flow Prediction Model Based on Improved GRU[J].,2023,33(07):208-214.[doi:10. 3969 / j. issn. 1673-629X. 2023. 07. 031]

点击复制

基于改进 GRU 的高速公路交通流量预测模型()

分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

33

期数:

2023年07期

页码:

208-214

栏目:

新型计算应用系统

出版日期:

2023-07-10

文章信息/Info

Title:

Highway Traffic Flow Prediction Model Based on Improved GRU

文章编号:

1673-629X(2023)07-0208-07

作者:

陈 刚

广西交通投资集团,广西 南宁 530022

Author(s):

CHEN Gang

Guangxi Communications Investment Group Corporation Ltd. ,Nanning 530022,China

关键词:

交通流量预测; 高速公路; 长短期记忆神经网络; 门控循环网络; 预测模型

Keywords:

traffic flow prediction; highway; long short-term memory neural network; gated recurrent unit; prediction mode

分类号:

TP183;U491

DOI:

10. 3969 / j. issn. 1673-629X. 2023. 07. 031

摘要:

交通流量预测常用于改善交通阻塞控制,是构建智能交通系统的基本技术之一。 为了能够准确、高精度预测高速公路交通

流量,充分利用交通流量的随机性、非线性及周期性等特征,设计了基于改进门控循环网络高速公路交通流预测模型,采用平均绝对误差(MAE)等 3 项评估指标,依托广西高速公路的交通流量数据集和芬兰公共交通管理局某自动测量站(LAM) 收集的道路交通信息上进行实例应用,并与长短期记忆神经网络( LSTM) 和门控循环网络( GRU) 进行对比。以芬兰数据集为例分析,在三层神经网络结构上采用 adam 优化器的改进 GRU 模型预测精度更高,其 RMSE 值为 8. 672 9,MAE 值为 6. 199 1, MAPE 值为 24. 76% ;在广西数据集上可得到同样的结论,三项指标均优于 LSTM 和 GRU。 进一步验证了所提模型的准确性和稳定性,以及在高速公路流量预测方面的实用性。

Abstract:

Traffic flow prediction is often used to improve traffic congestion control and is one of the basic techniques for building intelligent transportation systems. In order to predict highway traffic flow accurately and with high precision,and make?

full use of thecharacteristics of randomness,nonlinearity and periodicity of traffic flow,we design a highway traffic flow prediction model based on animproved gated recurrent network. Three evaluation indexes including mean absolute?

error ( MAE ) are applied, and an exampleapplication is carried out on the traffic flow data set of Guangxi highway?

and the road traffic information collected by an automatic measurement station ( LAM) of the Finnish Public Transport Administration, and comparison is made with the long short-term memory neuralnetwork ( LSTM) and gated recurrent unit ( GRU) . Using the Finnish dataset as an example analysis,the improved GRU model withadam optimizer on the three-layer neural network structure has higher prediction accuracy with RMSE value of 8.672 9,MAE value of 6.1991,and MAPE value of 24. 76% . The same conclusion can be obtained on the Guangxi dataset, and all three indexes are better thanLSTM and GRU. It further illustrates the accuracy and stability of the model proposed,and its usefulness in highway traffic prediction.

常用功能