The invention discloses an expressway short-time flow prediction method based on a self-supervised space-time Transform model. The expressway short-time flow prediction method comprises the following steps: S1, inputting traffic flow data; s2, data preprocessing; s3, setting model parameters; s4, a network model is constructed, and a self-supervised space-time Transform model composed of a space-time Transform module and a mask self-coding module is constructed; s5, training a network model, traversing each missing method and missing rate for training set data, and respectively introducing missing data to carry out model training; s6, the network model is tested, the trained network model is tested and verified through verification set data, and optimal model parameters are obtained. Through the prediction method, the space-time dependency in the traffic data can be better captured, the adaptability to missing data is enhanced through the model, and the prediction performance of the traffic flow is improved.
本发明的一种基于自监督时空Transformer模型的高速公路短时流量预测方法,包括以下步骤:S1.输入交通流量数据;S2.数据预处理;S3.模型参数设置;S4.构建网络模型,构建由时空Transformer模块、掩码自编码模块构成的自监督时空Transformer模型;S5.训练网络模型,对训练集数据遍历每种缺失方法和缺失率,分别引入缺失数据进行模型训练;S6.测试网络模型,将训练好的网络模型采用验证集数据进行测试和验证,得到最佳模型参数。通过本发明的预测方法能够更好地捕捉交通数据中的时空依赖性,通过本发明的模型增强了对缺失数据的适应能力,提高了交通流量的预测性能。
Expressway short-time flow prediction method based on self-supervised space-time Transform model
基于自监督时空Transformer模型的高速公路短时流量预测方法
15.10.2024
Patent
Elektronische Ressource
Chinesisch
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
Expressway short-time headstock prediction method based on trajectory data
Europäisches Patentamt | 2022
|Expressway vehicle space-time path prediction method and device
Europäisches Patentamt | 2025
|Expressway short-time vehicle speed prediction method considering dynamic traffic scene
Europäisches Patentamt | 2024
|An Expressway Short-Term Traffic Flow Prediction Model Based on Attention Mechanism
Springer Verlag | 2024
|