Wenquan Xu

School of Management Science and Engineering, Anhui University of Finance and Economics

Abstract：

The risk of global supply chain disruptions has become a core threat to business operations, making enhancing supply chain resilience a shared goal in academia and industry. This paper aims to build an early warning and emergency response system for supply chain disruption risks that integrates artificial intelligence technology. Through a systematic literature review, the study integrates multi-source heterogeneous data (operational data, environmental data, public opinion data). It uses principal component analysis (PCA) and expert scoring methods to construct a dynamic risk assessment index system covering three dimensions: supply side, operation side, and demand side. Furthermore, an LSTM-Transformer-Graph Neural Network (GNN) hybrid deep learning early warning model is designed to capture the temporal dependency, long-range correlation, and network conductivity of risks. At the emergency response level, a dynamic decision support module is developed by combining an improved genetic algorithm with a reinforcement learning framework, enabling real-time generation and optimization of emergency strategies. Finally, a prototype system is developed based on a microservices architecture, and case validations are conducted using three representative companies from the manufacturing, retail, and logistics industries. The empirical results show that the early warning model constructed in this study achieves an 89.6% prediction accuracy rate, a 12.3% improvement over the traditional BP neural network model. The emergency response module reduces the average decision-making time from 4 hours to 5 minutes, helps decrease supply chain recovery time by over 30%, and reduces disruption losses by 42%. This study not only provides practical intelligent tools for supply chain risk management but also offers an interdisciplinary research perspective and practical example for deepening supply chain resilience theory.

Key Words：

supply chain disruption risk; risk early warning; emergency response; Artificial Intelligence; supply chain resilience