A Deep Learning Approach for Network-wide Dynamic Traffic Prediction during Hurricane Evacuation
A Deep Learning Approach for Network-wide Dynamic Traffic Prediction during Hurricane Evacuation
复制标题
DOI：
10.1016/j.trc.2023.104126
复制
发表时间：
2022-02
期刊：
影响因子：
--
通讯作者：
Rezaur Rahman;Samiul Hasan
中科院分区：
文献类型：
--
作者：
Rezaur Rahman;Samiul Hasan
主动疏散交通管理在很大程度上依赖于高时空分辨率的交通流实时监测和预测。然而，由于预测飓风路径的突然变化以及由此导致的人口疏散行为的不确定性，疏散交通预测具有挑战性。此外，建模时空交通流模式需要更长的时间段内的大量数据，而疏散通常持续两到五天。在本文中，我们提出了一种新的数据驱动方法来预测网络规模的疏散交通。我们建立了一个动态图卷积长短期记忆神经网络（dgn - lstm）模型来学习飓风疏散过程中的网络动态。我们首先对非疏散期交通数据进行模型训练，发现该模型在预测非疏散期交通方面优于现有的深度学习模型，RMSE值为226.84。然而，将该模型应用于疏散交通预测时，RMSE值增加到1440.99。我们通过采用迁移学习方法来克服这一问题，该方法具有与疏散交通需求相关的附加特征，如疏散区域的距离、登陆时间和其他区域级特征，以控制从非疏散期到疏散期的信息传递（网络动态）。最终迁移学习DGCN-LSTM模型对疏散交通流的预测效果较好（RMSE = 399.69）。所实现的模型可用于较长时间（6小时以内）的疏散交通预测。它将协助运输机构启动适当的交通管理策略，以减少疏散交通的延误。
Proactive evacuation traffic management largely depends on real-time monitoring and prediction of traffic flow at a high spatiotemporal resolution. However, evacuation traffic prediction is challenging due to the uncertainties caused by sudden changes in projected hurricane paths and consequently populations’ evacuation behavior. Moreover, modeling spatiotemporal traffic flow patterns requires extensive data over a longer time period, whereas evacuations typically last for two to five days. In this paper, we present a novel data-driven approach for predicting evacuation traffic at a network scale. We develop a dynamic graph convolutional long short-term memory neural network (DGCN-LSTM) model to learn the network dynamics during hurricane evacuation. We first train the model for non-evacuation period traffic data and found that the model outperforms existing deep learning models for predicting non-evacuation period traffic with an RMSE value of 226.84. However, when the model is applied for predicting evacuation traffic, the RMSE value increased to 1440.99. We overcome this issue by adopting a transfer learning approach with additional features related to evacuation traffic demand such as distance from the evacuation zone, time to landfall, and other zonal level features to control the transfer of information (network dynamics) from non-evacuation periods to evacuation periods. The final transfer learned DGCN-LSTM model performs well to predict evacuation traffic flow (RMSE = 399.69). The implemented model can be applied to predict evacuation traffic over a longer forecasting horizon (up to 6-hour). It will assist transportation agencies to activate appropriate traffic management strategies to reduce delays for evacuating traffic.