A Hybrid Model of Interaction Between IoT and ERP Systems in the Context of Industry 4.0

Abstract

The interaction between Internet of Things (IoT) and Enterprise Resource Planning (ERP) systems in Industry 4.0 is still largely limited to data transfer and rarely turns sensor data into economically justified decisions. This paper proposes a hybrid model that combines edge-level preprocessing, a cloud probabilistic risk model, and an ERP decision layer governed by an operating-cost function, providing a systematic transition from raw IoT data to cost-optimal management actions. The model is validated in Python on two predictive-maintenance benchmarks - AI4I 2020 and the Microsoft Azure Predictive Maintenance dataset - using cross-validation and compared with a rule-based baseline and cloud-only classifiers (Logistic Regression and Random Forest). On strongly imbalanced data, the ERP cost-optimal threshold increases failure recall and reduces the total operating cost by 20-65% across both datasets relative to the cloud-only model, confirming the practical value of the approach.