Global supply chain disruptions, amplified by the COVID 19 pandemic and geopolitical tensions, have highlighted vulnerabilities in U.S. manufacturing systems, especially in sectors critical to national security and public health. To address these challenges, this study developed a simulation-based Artificial Intelligent (AI) driven resilience framework integrating predictive analytics, autonomous response systems, and digital thread architectures. A mixed methods approach was employed: qualitative insights were obtained through a systematic review of peer reviewed literature and policy reports, while quantitative findings were derived from simulation experiments using synthetic disruption datasets. The simulations demonstrated that Long Short-Term Memory (LSTM) networks outperformed Random Forest and XGBoost in disruption forecasting, achieving the highest predictive accuracy. Similarly, Deep Reinforcement Learning (DRL) simulations suggested a reduction in recovery time by nearly 40% and operational cost savings of approximately 18% when compared with baseline rule-based systems. Furthermore, simulated digital thread integration improved supply chain visibility, traceability, and coordination efficiency by over 25 points on average relative to traditional systems. While these outcomes were generated in a simulated environment rather than validated through real world datasets, the findings reinforced the feasibility and potential of AI driven resilience frameworks. The study contributes theoretically by conceptualizing resilience as a proactive, adaptive capability, and practically by providing evidence that such frameworks could support reshoring strategies, reduce foreign dependencies, and strengthen U.S. supply chain readiness.
Samuel Sunday Omotoso (Tue,) studied this question.