基于长时卷积深度网络的交通流预测方法

A Traffic Flow Prediction Method Based on Long-term Deep Convolution Network

准确地进行交通流预测对保障公共安全和解决交通拥堵具有重要的意义,是近些年来具有挑战性的课题之一。本研究提出了一种基于长时卷积深度网络的交通流预测方法,将交通网络进行格网划分,其中每个像素表示当前单元格内的流量,那么交通流预测的问题就可以转化为给定一个包含静态动画的序列,预测每个像素未来的运动。文中所提出的模型是基于深度学习中的卷积神经网络(CNN)和长短期记忆模型(LSTM),用CNN来捕捉交通流量的空间依赖性,用LSTM来捕捉交通流量的时间依赖性,最后再结合可视化工具,对隐藏单元进行分析,进一步解释模型的有效性。我们用深圳市和北京市的出租车轨迹数据进行实验,实验结果表明,长时卷积深度网络在精度和可解释性方面都有很好的表现。

Accurate traffic forecasting is of great significance for public safety and solving traffic congestion.It is one of the challenging topics in recent years.This study proposes a traffic flow prediction method based on the long-term Convolution depth network.The traffic network is divided into grids,where each pixel represents the current cell traffic,then the problem of traffic flow prediction can be translated into Set a sequence containing static animations to predict the future motion of each pixel.The models proposed in this paper are based on Convolutional Neural Network (CNN) and Long- and Short-Term Memory Model (LSTM) in deep learning.CNN is used to capture the spatial dependence of traffic flow,and LSTM is used to capture the time dependence of traffic flow.Finally,the Combined with visualization tools,the hidden Units are analyzed to further explain the effectiveness of the model.We use the taxi trajectory data from Shenzhen and Beijing to conduct experiments.The experimental results show that the long-term Convolution depth network has good performance in terms of accuracy and interpretability.