Scholarlink

<p>In the highly interconnected and dynamically evolving global supply chain, logistics disruption events show the characteristics of frequent occurrence, structural dependence and significant risk cascade, which brings severe challenges to prediction and mitigation decisions. Therefore, this paper proposes an intelligent expert system integrating graph structure risk inference and risk perception, which is used to predict and mitigate the logistics disruption in the global supply chain. Firstly, the multi-level supply chain graph model is constructed to explicitly describe the dependency between nodes and paths, and the risk propagation process at node level and network level is quantified through the graph structure driven risk inference mechanism; Then, the inferred structural risk characteristics are embedded into the multi model integrated prediction framework to realize the dynamic prediction of logistics interruption probability; On this basis, combined with expert rules and model reasoning results, an interpretable and executable mitigation decision scheme is generated. The evaluation is conducted on the experimental data set covering 36 months’ operation data, more than 500 supply chain nodes and more than 1000 logistics paths. The results show that compared with the prediction model without risk inference, the proposed method improves AUC by 5.3%-8.7%, reduces MAE by about 14.6%, and prolongs the average early warning time by 0.9–1.6 time units; In the high-risk scenario where multiple node failures and exogenous shocks are superimposed, the decline of system prediction performance is less than 20%, which is significantly better than the comparison method. At the same time, the mitigation decision support based on expert system can reduce the overall risk by about 30%-40% in high-risk scenarios. The above results verify the effectiveness, robustness and practical application value of the intelligent expert system in complex global supply chain logistics interruption prediction and mitigation decision-making.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An intelligent expert system for logistics disruption prediction and mitigation in global supply chains: Integrating graph-based risk inference with ensemble forecasting

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月16日

查看全文
  • Tianwen Zhao,
  • Guoqing Chen,
  • Piyapatr Busababodhin

摘要

In the highly interconnected and dynamically evolving global supply chain, logistics disruption events show the characteristics of frequent occurrence, structural dependence and significant risk cascade, which brings severe challenges to prediction and mitigation decisions. Therefore, this paper proposes an intelligent expert system integrating graph structure risk inference and risk perception, which is used to predict and mitigate the logistics disruption in the global supply chain. Firstly, the multi-level supply chain graph model is constructed to explicitly describe the dependency between nodes and paths, and the risk propagation process at node level and network level is quantified through the graph structure driven risk inference mechanism; Then, the inferred structural risk characteristics are embedded into the multi model integrated prediction framework to realize the dynamic prediction of logistics interruption probability; On this basis, combined with expert rules and model reasoning results, an interpretable and executable mitigation decision scheme is generated. The evaluation is conducted on the experimental data set covering 36 months’ operation data, more than 500 supply chain nodes and more than 1000 logistics paths. The results show that compared with the prediction model without risk inference, the proposed method improves AUC by 5.3%-8.7%, reduces MAE by about 14.6%, and prolongs the average early warning time by 0.9–1.6 time units; In the high-risk scenario where multiple node failures and exogenous shocks are superimposed, the decline of system prediction performance is less than 20%, which is significantly better than the comparison method. At the same time, the mitigation decision support based on expert system can reduce the overall risk by about 30%-40% in high-risk scenarios. The above results verify the effectiveness, robustness and practical application value of the intelligent expert system in complex global supply chain logistics interruption prediction and mitigation decision-making.