The invention discloses a multi-time-scale traffic flow speed prediction method based on vehicle-road cloud collaboration, and the method comprises the steps: carrying out the collection and data preprocessing of traffic flow data at a road end, and extracting traffic flow data xh, xd and xw of three time scales of each hour, daily and weekly; the cloud stores the preprocessed traffic flow data and dynamically updates weights wh, wd and ww; the vehicle end receives xh, xd and xw output by the road end and receives weights wh, wd and ww output by the cloud end, a plurality of space-time modules and a full connection layer are established for xh, xd and xw time scales respectively, a residual learning framework is adopted in each time scale, each space-time module comprises a space-time attention mechanism module and an LSTM-GCN module, and the LSTM-GCN module is connected with the vehicle end. And the space-time attention mechanism captures dynamic space-time correlation in the traffic flow data, the LSTM captures time features, the GCN captures space features, weighted fusion is carried out on the output of the full connection layer of each time scale, and the final predicted traffic flow speed # imgabs0 # is generated. The method is high in prediction accuracy.
本发明公开了一种基于车路云协同的多时间尺度交通流速度预测方法,路端进行交通流数据的采集与数据预处理,提取每时、每日、每周三个时间尺度的交通流数据χh、χd、χw;云端对预处理的交通流数据进行存储并动态更新权重wh、wd、ww;车端接收路端输出的χh、χd、χw,接收云端输出的权重wh、wd、ww,分别针对χh、χd、χw三个时间尺度建立多个时空模块和一个全连接层,每个时间尺度中采用残差学习框架,每个时空模块均包括一个时空注意力机制模块和一个LSTM‑GCN模块,时空注意力机制捕捉交通流数据中的动态时空相关性,LSTM捕捉时间特征,GCN捕捉空间特征,对每个时间尺度的全连接层的输出进行加权融合,生成最终的预测的交通流速度#imgabs0#本发明预测准确性高。
Multi-time scale traffic flow speed prediction method based on vehicle-road cloud cooperation
一种基于车路云协同的多时间尺度交通流速度预测方法
2024-08-06
Patent
Electronic Resource
Chinese
Traffic flow prediction system and method based on vehicle-road cooperation
European Patent Office | 2015
|European Patent Office | 2024
|Multi-scale traffic information publishing system and method based on vehicle-road cooperation
European Patent Office | 2020
|Road traffic operation situation multi-time scale prediction method
European Patent Office | 2020
|Traffic flow information acquisition device based on vehicle-road cooperation
European Patent Office | 2021
|