基于融合多源嵌入信息双向Mamba的提升电子商务物流效率研究
Enhancing E-Commerce Logistics Efficiency via Bi-Directional Mamba with Multi-Source Embedding Fusion
DOI: 10.12677/ecl.2026.15111226, PDF,    科研立项经费支持
作者: 王艺霖:浙江理工大学信息科学与工程学院,浙江 杭州
关键词: 电子商务物流;交通流预测;ETA估计;动态路径规划;双向Mamba;E-Commerce Logistics; Traffic Flow Forecasting; ETA Estimation; Dynamic Route Planning; Bi-Directional Mamba
摘要: 为提升电子商务物流的履约时效与仓配周转效率,关键在于增强路网速度与拥堵状态的可预见性,以支撑更准确的ETA估计、更优的动态路径规划与更提前的运力调度。针对现有交通流预测方法在长程依赖建模与动态空间相关刻画方面的不足,本文提出一种融合多源嵌入信息的双向Mamba交通流预测方法(MEBI-Mamba)。在时间维度,采用双向Mamba并行建模正向与逆向序列并融合,强化上下文表示;在空间维度,将路网节点按保持局部性的顺序序列化,利用空间Mamba捕获动态空间依赖。模型进一步融合时间、空间与自适应节点嵌入,并结合通道注意力与门控机制实现特征优化。在PEMS04与PEMS08上实验表明,所提方法在60分钟预测任务中MAE分别为7.55和6.39,RMSE分别为19.11和15.80,模型参数量仅为0.68 × 106个,推理速度超过400样本/秒。结果显示,该方法可为电商干支线与末端配送提供可靠预测支撑。
Abstract: To improve e-commerce logistics fulfillment timeliness and warehouse-distribution turnover efficiency, it is crucial to enhance the predictability of road-network speeds and congestion states, thereby enabling more accurate estimated time of arrival (ETA) estimation, better dynamic route planning, and earlier capacity dispatching. To address the limitations of existing traffic flow forecasting methods in modeling long-range dependencies and characterizing dynamic spatial correlations, this paper proposes a bi-directional Mamba-based traffic flow forecasting method with multi-source embedding fusion (MEBI-Mamba). In the temporal dimension, a bi-directional Mamba is adopted to model forward and backward sequences in parallel and fuse them to strengthen contextual representations. In the spatial dimension, road-network nodes are serialized in a locality-preserving order, and a spatial Mamba is used to capture dynamic spatial dependencies. The model further integrates temporal, spatial, and adaptive node embeddings, and employs channel attention and gating mechanisms for feature refinement. Experiments on PEMS04 and PEMS08 demonstrate that, for the 60-minute forecasting task, the proposed method achieves MAE values of 7.55 and 6.39 and RMSE values of 19.11 and 15.80, respectively, with only 0.68 × 106 parameters and an inference speed exceeding 400 samples/s. The results indicate that the proposed method can provide reliable predictive support for e-commerce trunk-line and last-mile delivery.
文章引用:王艺霖. 基于融合多源嵌入信息双向Mamba的提升电子商务物流效率研究[J]. 电子商务评论, 2026, 15(11): 185-194. https://doi.org/10.12677/ecl.2026.15111226

参考文献

[1] 张世宇, 叶纯, 代丽. 农村电商物流的研究现状及热点分析——基于CiteSpace知识图谱分析[J]. 电子商务评论, 2025, 14(12): 7185-7196.
[2] 张玲君. 面向智慧养老社区的O2O电商服务整合模式与优化策略研究[J]. 老龄化研究, 2026, 13(2): 195-201.
https://doi.org/10.12677/ar.2026.132022
[3] Yin, X., Wu, G., Wei, J., Shen, Y., Qi, H. and Yin, B. (2022) Deep Learning on Traffic Prediction: Methods, Analysis, and Future Directions. IEEE Transactions on Intelligent Transportation Systems, 23, 4927-4943.
https://doi.org/10.1109/tits.2021.3054840
[4] Dey, R. and Salem, F.M. (2017) Gate-Variants of Gated Recurrent Unit (GRU) Neural Networks. 2017 IEEE 60th International Midwest Symposium on Circuits and Systems (MWSCAS), Boston, 6-9 August 2017, 1597-1600.
https://doi.org/10.1109/mwscas.2017.8053243
[5] Yu, B., Yin, H. and Zhu, Z. (2017) Spatio-Temporal Graph Convolutional Networks: A Deep Learning Framework for Traffic Forecasting.
https://arxiv.org/abs/1709.04875
[6] Li, Y., Yu, R., Shahabi, C., et al. (2018) Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting. Vancouver: International Conference on Learning Representations, Vancouver, 2018, 1-16.
[7] Xu, M., Dai, W., Liu, C., et al. (2022) STAformer: Spatial-Temporal Attention Transformer for Traffic Flow Forecasting. Orlando: 2022 IEEE International Conference on Data Mining, Orlando, 2022, 1345-1350.
[8] Jiang, J., Han, C., Zhao, W., et al. (2023) PDFormer: Propagation Delay-Aware Dynamic Long-Range Transformer for Traffic Flow Prediction. arXiv:2301.07945.
https://arxiv.org/abs/2301.07945
[9] Awad, M. and Khanna, R. (2015) Support Vector Regression. In: Awad, M. and Khanna, R., Eds., Efficient Learning Machines, Apress, 67-80.
https://doi.org/10.1007/978-1-4302-5990-9_4
[10] Guo, S., Lin, Y., Feng, N., Song, C. and Wan, H. (2019) Attention Based Spatial-Temporal Graph Convolutional Networks for Traffic Flow Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence, 33, 922-929.
https://doi.org/10.1609/aaai.v33i01.3301922
[11] Lin, W., Zhang, Z., Ren, G., Zhao, Y., Ma, J. and Cao, Q. (2025) MGCN: Mamba-Integrated Spatiotemporal Graph Convolutional Network for Long-Term Traffic Forecasting. Knowledge-Based Systems, 309, Article 112875.
https://doi.org/10.1016/j.knosys.2024.112875
[12] Yuan, D., Xue, J., Su, J., et al. (2024) ST-Mamba: Spatial-Temporal Mamba for Traffic Flow Estimation Recovery Using Limited Data. arXiv:2407.08558.
https://arxiv.org/abs/2407.08558