无线电工程2026，Vol.56Issue(1)：70-82,13.DOI:10.3969/j.issn.1003-3106.2026.01.008
基于IVMD与分层自适应图卷积的城市交通流量预测
Urban Traffic Flow Prediction Based on IVMD and Hierarchical Adaptive Graph Convolution
摘要
Abstract
To solve the problem of non-stationarity and high dynamic correlation of traffic flow sequence,which makes it difficult to predict traffic flow,a new traffic flow prediction model based on graph convolutional neural network is proposed.In the model design,an Improved Variational Mode Decomposition(IVMD)module is established with the help of particle swarm optimization algorithm to decompose the original traffic flow sequence into a series of modal function components and thus alleviating the non-stationarity of the traffic flow sequence;three Gated Fusion Network(GFN)are utilized to cross-fuse the historical traffic flow,Point of Interest(POI),and time-stamped data to learn the spatio-temporal interaction features;a Multi-scale Decomposable Mixing(MDM)module is utilized to capture short-term fluctuations and long-term trend of traffic flow,and extract seasonal and trend features of traffic flow;meanwhile,dynamic spatio-temporal features of traffic flow are captured by constructing a hierarchical spatio-temporal convolutional block to solve the dynamic correlation problem among traffic nodes;finally,advanced Hierarchical Spatial-Temporal Convolution Block(HST-Conv Block)features are obtained by summing up individual features through the summation module to enhance the traffic flow prediction performance of the model.Comparative numerical experiments show that the obtained model has strong feature extraction capability and can effectively improve the accuracy of traffic flow prediction.关键词
交通流量预测/变分模态分解/分层自适应图卷积/兴趣点Key words
traffic flow prediction/variational mode decomposition/hierarchical adaptive graph convolution/POI分类
信息技术与安全科学引用本文复制引用
葛洪霞,张著洪..基于IVMD与分层自适应图卷积的城市交通流量预测[J].无线电工程,2026,56(1):70-82,13.基金项目
国家自然科学基金(62063002)National Natural Science Foundation of China(62063002) （62063002）
