The invention provides a traffic large model training method and device, relates to the technical field of artificial intelligence, in particular to the technical field of deep learning, and can be applied to the field of intelligent traffic. According to one specific embodiment, the method comprises the steps that a training sample is acquired, and the training sample comprises first historical traffic data and a first future transit time true value; performing feature extraction on the first historical traffic data to obtain a first historical traffic feature; inputting the first historical traffic feature into a converter model to obtain a first future transit time predicted value; calculating a first loss based on the first future transit time predicted value and the first future transit time true value; and adjusting parameters of the converter model based on the first loss to obtain a traffic large model. The traffic large model can be used for predicting the future passing time, and the prediction accuracy of the passing time is improved.
本公开提供了一种交通大模型训练方法和装置,涉及人工智能技术领域,具体为深度学习技术领域,可应用于智能交通领域。该方法的一具体实施方式包括:获取训练样本,其中,训练样本包括第一历史交通数据和第一未来通行时间真实值;对第一历史交通数据进行特征提取,得到第一历史交通特征;将第一历史交通特征输入至转换器模型,得到第一未来通行时间预测值;基于第一未来通行时间预测值和第一未来通行时间真实值,计算第一损失;基于第一损失调整转换器模型的参数,得到交通大模型。该实施方式交通大模型可以用于预测未来通行时间,提高了通行时间的预测准确度。
Traffic large model training method and device
交通大模型训练方法和装置
2024-09-24
Patent
Electronic Resource
Chinese
IPC: | G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
Model training method, traffic flow prediction method and device
European Patent Office | 2024
|Traffic flow prediction model training method, traffic flow prediction method, device and equipment
European Patent Office | 2023
|Traffic track prediction model training method, traffic track prediction method and related device
European Patent Office | 2024
|European Patent Office | 2024
|European Patent Office | 2016
|