湖北汽车工业学院学报2026，Vol.40Issue(1)：12-17,29,7.DOI:10.3969/j.issn.1008-5483.2026.01.003
基于混合深度学习的网络流量预测模型研究
Research on Network Traffic Prediction Model Based on Hybrid Deep Learning
摘要
Abstract
Aiming at the shortcomings of existing network traffic prediction models,such as single input data,difficult parameter optimization,and low prediction accuracy,a hybrid prediction model was pro-posed,which incorporated variational mode decomposition,a Transformer encoder,and a bidirec-tional gated recurrent unit(BiGRU),and an improved sparrow search algorithm was used to optimize the mod-el parameters.The model extracted multi-scale frequency characteristics of the traffic se-quence and introduced a bidirectional information interaction mechanism between the Transformer encoder and the BiGRU module to further improve the prediction performance through feature feedback and fusion.The experimental results show that the evaluation indicators of the proposed model reduce by more than 60%compared with other methods.In addition,ablation experiments verify the effectiveness of each module in improving prediction accuracy.关键词
网络流量预测/变分模态分解/Transformer编码器/双向GRU/信息交换机制/改进麻雀搜索算法Key words
network traffic prediction/variational mode decomposition/Transformer encoder/BiGRU/information ex-change mechanism/improved sparrow search algorithm分类
信息技术与安全科学引用本文复制引用
张笑航,翟亚红,徐龙艳,范志远..基于混合深度学习的网络流量预测模型研究[J].湖北汽车工业学院学报,2026,40(1):12-17,29,7.基金项目
湖北省教育厅科学技术研究计划项目(D202111802) （D202111802）
湖北省重点研发计划项目(2022BEC008) （2022BEC008）
