摘要
随着深度学习的发展,交通流预测的准确率越发提高,对时间序列的交通流预测进行研究,基于一种双阶段注意力机制循环神经网络模型(DA-RNN),以解决当前在交通流量的时间序列预测中存在的难以捕捉时间数据序列之间的相关性导致预测不够准确的问题,并解决实验中存在的过拟合现象。论文基于PEMS04数据进行实验并将预测结果与LSTM、GRU模型的预测结果进行对比,表明该时序预测模型具有良好的性能,可为交通管理与控制提供有效依据。
With the development of deep learning,the accuracy of traffic flow forecasting is increasing.This article starts from the perspective of time series forecasting traffic flow,and is based on a two-stage attention mechanism recurrent neural network model(DA-RNN)to solve the current traffic flow.It is difficult to capture the correlation between time data series in the time series forecasting of traffic,which leads to the problem of inaccurate prediction and solves the problem of overfitting in the experiment.This paper conducts experiments based on PEMS04 data and compares the prediction results with the prediction results of LSTM and GRU models.It shows that this time series prediction model has good performance and can provide an effective basis for traffic management and control.
作者
王健
王峥
WANG Jian;WANG Zheng(Wuhan Research Institute of Posts and Telecommunications,Wuhan 430074;Nanjing Fenghuo Tiandi Communication Technology Co.,Ltd.,Nanjing 210019)
出处
《计算机与数字工程》
2024年第4期1251-1256,共6页
Computer & Digital Engineering
