The invention discloses an expressway congestion dissipation time prediction method based on deep meta-learning, and the method comprises the following steps: collecting vehicle trajectory data of a congested road section and upstream and downstream ETC data, extracting vehicle and traffic flow features, collecting abnormal event information of the road section, and arranging the abnormal event information to form a data set; capturing spatial features of the traffic flow from the data set by using CNN, and capturing time features of the traffic flow from the data set by using GRU; constructing a congestion dissipation time prediction model; and for different types of abnormal events, the congestion dissipation time prediction model adopts an MAML method to perform task division learning, so that congestion dissipation time prediction results under different types of abnormal events are obtained. The method provided by the invention can effectively overcome the limitation of an existing congestion dissipation time prediction method in dealing with an abnormal event, and realizes higher prediction precision and adaptability.
本发明公开了一种基于深度元学习的高速公路拥堵消散时间预测方法,包括以下步骤:采集拥堵路段的车辆轨迹数据以及上下游的ETC数据,提取车辆与交通流特征,收集路段异常事件信息整理形成数据集;使用CNN从数据集中捕获交通流的空间特征,使用GRU从数据集中捕获交通流的时间特征;构建拥堵消散时间预测模型;针对不同类型的异常事件,拥堵消散时间预测模型采用MAML方法进行分任务学习,从而得到不同异常事件类型下的拥堵消散时间预测结果。本发明方法能够有效克服现有拥堵消散时间预测方法在应对异常事件时存在的局限性,实现更高的预测精度和适应性。
Highway congestion dissipation time prediction method based on deep meta-learning
一种基于深度元学习的高速公路拥堵消散时间预测方法
2023-11-07
Patent
Electronic Resource
Chinese
Expressway congestion dissipation time prediction method considering traffic event information
European Patent Office | 2024
|Highway short-time traffic flow prediction method based on deep learning
European Patent Office | 2023
|Online Contents | 1995
|Highway congestion prediction method and system based on ETC portal frame
European Patent Office | 2020
|ETC portal frame-based highway congestion 5G network prediction system
European Patent Office | 2021
|