This study reinterprets the supply chain “ripple effect” as a destabilizing positive feedback loop and proposes a cybernetic control mechanism to restore homeostasis. Moving beyond reactive monitoring, it aims to develop a predictive Digital Twin framework that functions as a “variety attenuator” (per Ashby's Law) to regulate delivery uncertainty in global logistics networks.
Adopting a dual-stage hybrid methodology, the study first maps the intellectual boundaries of the domain via a systematic bibliometric analysis (n = 267). Subsequently, it constructs a predictive control engine using a large-scale real-world logistics dataset (n = 180,519). Three machine learning architectures (Naïve Bayes, Artificial Neural Network, Random Forest) are benchmarked to test their capacity to handle environmental variety.
Experimental results demonstrate that the Random Forest ensemble acts as the most effective variety attenuator, achieving 90.90% prediction accuracy. Crucially, feature importance analysis reveals that delays are structurally determined by “Shipping Mode” and “Geographical Destination,” rather than being purely stochastic. This enables the design of negative feedback mechanisms for system stability.
This research bridges the gap between engineering-focused Digital Twins and systems theory. By forecasting operational delays at the pre-shipment stage, the framework targets the genesis of disruptions rather than their macroscopic propagation. It operationalizes cybernetic concepts by validating a data-driven intelligence function (System 4) that empowers managers to preempt the ripple effect and implement cost-effective resilience through selective intervention.
