文献类型：期刊文献

中文题名：基于双尺度自适应令牌注意力的交通流量预测

英文题名：Traffic Flow Prediction Based on Dual-Scale Adaptive Token Attention

作者：郭津延[1];郑红[1];杜佳宇[1];罗俞建[1];李鹏威[1];单蓉胜[2]

机构：[1]华东理工大学信息科学与工程学院,上海200237;[2]上海交通大学网络空间安全学院,上海200240

年份：2025

卷号：51

期号：6

起止页码：817

中文期刊名：华东理工大学学报(自然科学版)

外文期刊名：Journal of East China University of Science and Technology

收录：;北大核心:【北大核心2023】；

基金：上海市2024年度“科技创新行动计划”资助(24BC3200500,24BC3200300)。

语种：中文

中文关键词：智能交通;交通流预测;注意力机制;Transformer;深度学习

外文关键词：intelligent transportation;traffic flow prediction;attention mechanism;Transformer;deep learning

摘要：针对现有交通流量预测方法存在计算复杂度高、实时性差以及局部与全局特征整合不足等缺点,本文提出了一种基于双尺度自适应令牌注意力的交通流量预测模型。该模型结合双尺度自适应令牌注意力机制提取复杂时空特征:通过双尺度可学习池化得到的令牌分别捕获数据的长期和短期特征,并利用自适应令牌注意力机制整合全局依赖关系,提升预测准确性和效率。在两个公开数据集上进行实验验证,结果表明该方法在预测精度和计算效率上优于现有主流模型,适用于实时交通流量预测场景,为智能交通系统提供了一种高效、精准的解决方案。
To address the shortcomings of existing methods in traffic flow prediction concerning computational complexity,real-time performance,and the integration of local and global features,this paper proposes a traffic flow prediction model that employs dual-scale adaptive token attention.The model incorporates a dual-scale adaptive token attention mechanism designed to extract complex spatio-temporal features while optimizing computational efficiency.Through dual-scale learnable pooling operations,the resulting tokens effectively capture both long-term and short-term temporal features of the data.Furthermore,the adaptive token attention mechanism integrates global dependencies to enhance prediction accuracy and operational efficiency.Experimental results on two public datasets demonstrates that the proposed method outperforms existing mainstream models in both prediction accuracy and computational efficiency.Particularly suitable for real-time traffic flow prediction scenarios,this approach provides an efficient and accurate solution for intelligent transportation systems,exhibiting significant theoretical and practical implications.