<![CDATA[This study evaluates an integrated PLC–SCADA–IIoT architecture with edge analytics and a closed-loop digital twin to realize a smart factory/CPPS in a brownfield environment. A design–build–measure–learn methodology is applied with interoperability standards (OPC UA/MQTT), network segmentation (optional TSN), and security governance based on IEC 62443. A hybrid (physics-informed + data-driven) predictive model is built and deployed at the edge; the digital twin is synchronized through state estimation (Δt ≈ 100–200 ms) to provide set-point recommendations and maintenance scheduling audited by the PLC safety guard. Factorial tests on combinations of line speed and degradation levels show significant performance improvements (Welch t-test, α=0.05): OEE increases by ~6–12 absolute points, cycle time decreases by 5–8%, energy/unit decreases by ~8%, and scrap decreases by ~35–40% at severe degradation. The increase in determinism is reflected in the decrease in p95 latency (PLC↔Edge 55→18 ms, Edge↔SCADA 85→35 ms). In predictive maintenance, performance improved (AUROC 0.78→0.94; PR-AUC 0.41→0.72; F1 0.56→0.78; RUL-MAE 22.1→9.4 hours) with the false alarm rate decreasing by 0.095→0.038. The safety posture increased from 1.2–2.0 to 3.9–4.5 (scale 0–5). The study’s key contributions are the co-design of a PLC deterministic loop with an AI adaptive loop at the edge, a closed-loop hybrid digital twin, and the measurement of technical–business benefits linked to ROI, providing a practical, incremental adoption path for manufacturers—especially SMEs—towards smart factories.]]>
Copyrights © 2024
