Stochastic Multi-Tier Supply Chain Resilience Optimization via Adaptive Dynamic Programming and Bayesian Demand Signal Decomposition

Abstract

Contemporary global supply chains operate under compounding systemic disruptions, rendering traditional deterministic optimization frameworks structurally inadequate. This study proposes a hybrid stochastic optimization model integrating Adaptive Dynamic Programming (ADP) with Bayesian demand signal decomposition to enhance multi-tier supply chain resilience under uncertainty. Empirical validation is conducted using longitudinal operational datasets from manufacturing firms across three industry verticals over a 48-month horizon. The model demonstrates a statistically significant 23.7% improvement in disruption recovery velocity and a 17.4% reduction in total landed cost variance compared to conventional safety-stock and scenario-planning benchmarks. Furthermore, the framework introduces a novel resilience elasticity coefficient enabling real-time reconfiguration of sourcing node hierarchies. Findings carry substantive implications for operations executives and strategic procurement leaders navigating post-pandemic and geopolitical volatility contexts.