Acta Logistica | 2026
Authors: Hermans M.; Tamas P.
DOI: 10.22306/al.v13i1.745
Journal: Acta Logistica
Year: 2026
Publisher: 4S go, s.r.o
Document Type: Article
Open Access: All Open Access; Gold Open Access
Cited by: 0
In maintenance logistics, optimising material and information flows requires predictive methodologies that cope with uncertainty and support decision-making across the maintenance supply chain. This paper proposes a Bayesian Neural Network (BNN) approach to predict maintenance demand and improve logistical planning of spare parts and technician deployment in flow-based production systems. By incorporating uncertainty directly into model outputs, the BNN provides probabilistic predictions that enhance robustness, precision, and responsiveness within logistics operations. The model is trained and validated on historical maintenance data and synthetic scenarios generated to reflect stochastic degradation and repair behaviour typical of industrial flow environments. Compared to deterministic benchmarks, the proposed approach demonstrates superior logistics performance, with improvements in downtime reduction, spare-parts availability, and overall material-flow efficiency. These insights highlight the benefit of probabilistic modelling as a decision-support tool for maintenance logistics in complex manufacturing logistics systems. © 2026, 4S go, s.r.o. All rights reserved.
Bayesian Neural Networks; maintenance logistics; material flow systems; predictive maintenance; supply chain planning