The invention discloses a traffic flow measurement method based on road network priori knowledge and a mask auto-encoder, which comprises the following steps of: dividing a target area into grids and constructing a traffic flow matrix corresponding to the grids, then partitioning to obtain a data block sequence, constructing map data of the target area to obtain a map block sequence, and constructing a mask auto-encoder according to the map block sequence; constructing and training a traffic flow inference model based on a mask auto-encoder, performing mask processing on the data block based on the self-attention score, and replacing the mask data block with a corresponding map block for inferring the traffic flow of the mask data block; and selecting a part of grid blocks in a target area to deploy traffic flow sensors, constructing a data block sequence by using the actually measured traffic flow of the part of grid blocks and the map blocks of the mask grid blocks, and inputting the data block sequence into the trained traffic flow inference model to obtain a mask grid block traffic flow inference result, thereby completing measurement. According to the invention, the traffic flow of each grid in the whole target area can be obtained by means of the traffic flow inference model, and hardware resources are saved.
本发明公开了一种基于路网先验知识和掩码自编码器的交通流量测量方法,将目标区域划分为网格并构建网格对应的交通流量矩阵,然后分块得到数据块序列,对目标区域的地图数据构建得到地图块序列,构建并训练基于掩码自编码器的交通流量推断模型,基于自注意力分数对数据块进行掩码处理,采用对应的地图块代替掩码数据块用于推断掩码数据块的交通流量;在目标区域中选择部分网格块部署交通流量传感器,将这部分网格块的实际测量交通流量和掩码网格块的地图块一起构建数据块序列,输入训练好的交通流量推断模型得到掩码网格块交通流量推断结果,从而完成测量。本发明可以借助交通流量推断模型获取整个目标区域内各个网格的交通流量,节约硬件资源。
Traffic flow measurement method based on road network prior knowledge and mask auto-encoder
基于路网先验知识和掩码自编码器的交通流量测量方法
2024-10-22
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen |
Traffic flow prediction method for space-time diagram auto-encoder
European Patent Office | 2023
|Traffic anomaly detection method based on graph convolutional neural network auto-encoder
European Patent Office | 2023
|Road network dynamic traffic distribution method based on Encoder-Decoder deep neural network
European Patent Office | 2021
|