<![CDATA[Cyber-physical systems (CPS) are increasingly playing a key role in intelligent decision support on the shop floor, as they enable the integration of the physical production assets with sensing, communication, digital modelling, analytics, and control. This review summarizes the modelling techniques in CPS for responsive and adaptive manufacturing, focusing on the architectures, data integration, real-time data analytics, decision logic, synchronization, and operational applications. The reviewed evidence indicates that CPS is able to assist in dynamic production scheduling, machine monitoring, fault diagnosis, material-flow optimization, human–machine collaboration, energy management, and process control. The technologies above are complementary, but their value relies on data accuracy, low latency communications, interoperability, and feedback. The current challenges within the data quality and availability, complexity of models, scalability, protocol heterogeneity, real-time implementation, cyber security, privacy, human-centred design are significant limitations. The review also highlights the lack of a seamless transition from the development of technically sophisticated prototypes to reliable industrial implementation, especially in small and medium manufacturing sectors, and those with legacy equipment and data systems. Implementing explainable and adaptive AI, coordinating with edge and cloud workloads, building interlinked digital twin ecosystems, ensuring secure middleware, establishing a standard for data models, and self-optimizing control with human oversight should be the focus of future progress. CPS modelling should be considered as a socio-technical decision architecture and not just as a set of standalone technologies. The industrial value will rely on the balance between modelling fidelity, computation speed, interoperability, affordability, security, shop floor operator trust and measurable shop floor performance. This synthesis elucidates the conditions necessary to ensure reliable industrial implementation.]]>
Copyrights © 2026
