Toward Viable Agricultural Supply Chains: A Digital Twin Framework with Artificial Intelligence and Immune-Inspired Control

This working paper proposes an integrated framework that combines digital twin technology, artificial intelligence, immune-inspired regulation, and adaptive memory to support viability-oriented decision-making in agricultural supply chains under disruption. The study is motivated by the growing interest in digital twins in both supply chain and agricultural research, alongside the limited development of frameworks capable of moving beyond monitoring and prediction toward dynamic regulation and learning. In response to this gap, the proposed architecture is structured around four tightly coupled components: a digital twin as the cyber-physical representation layer, an AI-driven state estimation and predictive modeling module, the Supply Chain Immune System (SCIS) as the regulatory layer, and Immune-Structural Adaptive Response (RAIE) as the adaptive memory layer. These components are formalized through a dynamic system in which performance evolves according to disruption effects, corrective actions, and accumulated experience. To illustrate the behavior of the framework, simulation experiments were conducted under baseline, single-disruption, and repeated-disruption scenarios. The results show that predictive capabilities improve anticipatory response, but their effect remains limited when not supported by adaptive regulation. In contrast, the integration of SCIS and RAIE leads to faster recovery, lower performance degradation, and more stable behavior under recurrent disturbances. The findings suggest that viability in agricultural supply chains depends not only on visibility and prediction, but also on the coordinated interaction of representation, control, and learning. The study contributes a conceptual and computational foundation for advancing digital twins in agriculture toward adaptive, disruption-aware, and viability-oriented systems.