| Title |
A Study on Physical AI Systems Based on PLC?AI Collaboration in Digital Twin Environments |
| Authors |
Seung-Hoon Kwon ; Eun-Hyeok Choi |
| DOI |
https://doi.org/10.5207/JIEIE.2026.40.3.204 |
| Keywords |
Automation equipment; Digital twin; OPC UA; Physical AI; Smart factory |
| Abstract |
This paper proposes a system architecture for implementing Physical AI in digital twin?based automation systems and experimentally evaluates its performance using a multi-process automated system (MPS). The proposed architecture separates rule-based safety control executed by a PLC from AI-driven decision-making, enabling collaborative operation within a closed-loop control framework. A digital twin environment is employed to generate synthetic data and optimize control parameters, while an IFM (Industry Foundation Module)?based learning structure enables continuous model adaptation using both real and virtual data. Experimental results demonstrate that the proposed system maintains production cycle time within ±10% of the target under varying process conditions, and achieves steady-state control performance with an average deviation of approximately ±5 seconds. The proposed system integrates Digital Twin-based simulation, IFM-based AI learning, and OPC UA-based closed-loop control to enable adaptive autonomous manufacturing in multi-process automation environments. These results validate the feasibility of applying Physical AI to multi-process automation environments and provide practical insights into system architecture and data flow design for autonomous manufacturing systems. |