The invention discloses a traffic road network coding representation learning method based on a mask graph attention mechanism, which utilizes the context information extraction capability of a BERT model to enlarge the adjacent space information extraction order of the graph attention mechanism, so that the model can complete the deep representation of road network traffic flow data and effectively extract the road network space characteristics. According to the RNERT model, a mask pre-training algorithm is adopted, and the generality capability of road network spatial feature extraction is learned. For the missing data completion task, the RNERT pre-training model is directly used to carry out completion calculation; for road network traffic flow prediction, space-time multi-dimensional features are extracted based on a traffic road network gated loop network (GRU-RNERT) represented by graph attention coding.
本发明公开了一种基于掩码图注意力机制的交通路网编码表征学习方法,利用BERT模型的上下文信息提取能力,扩大图注意力机制的邻接空间信息提取阶数,使模型能完成对路网交通流数据的深度表征,对路网空间特征进行有效提取。所述RNERT模型采用掩码(mask)预训练算法,学习路网空间特征提取的共性能力。对于缺失数据补全任务直接使用RNERT预训练模型进行补全计算;对于路网交通流预测,基于图注意力编码表示的交通路网门控循环网络(GRU‑RNERT),对其中的时空多维度特征进行提取。
Traffic network coding representation learning method based on mask pattern attention mechanism
一种基于掩码图注意力机制的交通路网编码表征学习方法
2024-03-12
Patent
Electronic Resource
Chinese
Deep reinforcement learning traffic signal control method based on attention mechanism
European Patent Office | 2024
|Personalized federated learning traffic flow prediction method based on prior attention mechanism
European Patent Office | 2024
|Deep reinforcement learning traffic signal decision system and method based on attention mechanism
European Patent Office | 2024
|Traffic flow prediction method based on attention mechanism
European Patent Office | 2022
|Dynamic space-time neural network traffic flow prediction method based on attention mechanism
European Patent Office | 2023
|