随着全球经济与市场环境变革，不确定因素产生的影响攀升，对供应链的决策产生了新的挑战。企业需要构建具有弹性供应链以应对不断变化的需求，增强抗风险能力。成熟的供应链能够敏捷地感知外部环境和自身动态，通过透明的数据系统与持续的决策技术进行不断调整，将不确定性转化为成长机遇。因此，本文针对企业订单从发货到交付的履约过程，基于动态更新的数据监控系统和管理各种共享运力资源的网络货运平台，对在线产生的订单实时规划运输路径，在承诺时效、限量资源和节点容量的约束下，实现成本、时效和资源利用率三种目标之间的最优平衡。本文设计了四种在线策略用于在线订单的路由规划与分配。首先是贪婪策略，为每个订单分配多目标最优路径，这种方法结构简洁，能在短时间内快速响应在线请求。其次是模仿策略，通过统计离线样本的最优决策，为在线到达的相似订单直接分配样本最优路径。此外，本文提出一种缓冲策略，通过缓冲池积累在线订单，当累积一定数量后，统一对缓冲池中的订单采用启发式算法求解。最后，本文还设计了一种动态调整策略，在订单完成运输路径匹配后，在其到达的路径节点处重新优化，提高决策系统的灵活性与适应力。通过真实的物流网络与企业案例，本文对比并评估了不同在线策略的求解效果，讨论了不同参数和数据分布对策略造成的影响。实验结果表明，结合缓冲策略和动态调整策略的集成方案在不同规模的问题上展现出良好的性能，同时能够适应不同的需求分布，具有鲁棒性和应用前景，有助于帮助企业建设持续决策的弹性供应链，在不确定的市场环境中保持竞争力。

With the rapid development of e-commerce and the proliferation of mobile devices, new retail formats such as community group buying and live streaming e-commerce have emerged, offering brand manufacturers the opportunity to transform their sales channels. The shift from traditional single offline retail channels to the parallel development of online and offline multi-channel channels has led to the era of full-channel integration. However, the cross-channel supply chain under full-channel transformation has posed new challenges for the delivery of enterprise orders, highlighting the need for flexible delivery systems that can respond to changing demands and enhance risk resistance.This paper proposes a delivery decision-making system that can perceive the external environment and its own dynamics in an agile manner, continuously adjusting through transparent data systems and decision-making techniques to turn uncertainty into growth opportunities. The proposed system focuses on the fulfillment process of enterprise orders, from shipment to delivery, and plans transportation paths for orders generated online in real-time, to achieve an optimal balance between cost, timeliness, and resource utilization under the constraints of committed timeliness, limited resources, and node capacity.The paper proposes four online strategies for routing and allocation of online orders, namely the greedy strategy, imitation strategy, buffering strategy, and dynamic adjustment strategy. Through real logistics networks and enterprise cases, the paper compares and evaluates the effectiveness of different online strategies, and discusses the impact of different parameters and data distributions on the strategies. The experimental results show that the integrated solution combining buffering and dynamic adjustment strategies exhibits good performance on problems of different scales, and can adapt to different demand distributions with robustness and application prospects. This system can help enterprises build resilient supply chains for continuous decision-making and maintain competitiveness in uncertain market environments.