ML-Enhanced Simulation for Industry 4.0: Integrating Heterogeneous Systems via Communication Infrastructure

Abstract

Industry 4.0 depends on the ability to connect heterogeneous systems, yet students and practitioners rarely have a low-risk environment in which to practice this kind of systems integration. This article presents a virtual-prototyping architecture developed and tested, linking a discrete-event simulation tool with external machine learning models through industrial communication protocols. The resulting artifacts and method are a contribution to systems engineering education, practice, and development. A two-phase empirical virtual-prototyping approach was used. First, Open Platform Communications Unified Architecture and Message Queuing Telemetry Transport were prototyped and compared as communication layers between Siemens Tecnomatix Plant Simulation and Python-based machine learning clients. Second, for this project, the more suitable protocol was applied to three increasingly complex use cases, addressing automated guided vehicle capacity, conveyor speed control, and process bottleneck identification. The use cases were assessed against the Technology Readiness Level scale. Open Platform Communications Unified Architecture provided reliable, real-time, bidirectional data exchange, while Message Queuing Telemetry Transport proved less stable for this application. The three use cases each demonstrated feasible simulation-machine learning integration, and the overall prototype reached Technology Readiness Level 4. Beyond its contribution to I4.0 practice, the staged research design, the use of Technology Readiness Levels as a maturity and reflection instrument, and the low-cost, risk-free nature of virtual prototyping constitute a transferable pedagogical pattern for systems engineering curricula, capstone projects, and competency-based training.