Design and Implementation of a Machine Feedback System with Dual Digital Twins for Industrial Safety Applications

The industry 4.0 paradigm ensures that modern machinery is networked and equipped with sensors capable of sharing operational states with the digital factory. This work pursues two main goals: assessing the system's capability to manage hazard notifications within defined uncertainty bounds and designing a prototype of an ïntelligent" machine system. Activities were divided into two main areas. First, a communication system was developed and implemented to enable the real machine to exchange operational data with a Python-based backend server. The Data Collector Manager was configured to define the machine network, acquisition parameters, and communication protocols, ensuring reliable machine-server interactions with metrological traceability. Data were acquired via a real CN industrial network protocol, allowing efficient transmission of operational information with adequate temporal resolution and management of downtime and maintenance events. Second, a Python-based digital twins were deployed on Raspberry Pi devices to emulate machine-server communication. The digital twin includes an interface allowing users to simulate machine states, alarms, and maintenance actions, transmitted to the same backend server used by the real machine. This guarantees synchronized and coherent interaction between physical and virtual systems, supported by precise time stamping. A unified dashboard, hosted on an external PC, enables real-time monitoring with minimal latency of machine status, diagnostics, and alarms from both real and virtual systems. The system successfully identified safety violations, such as spindle guard tampering, through predefined alarm thresholds with high sensitivity and specificity, ensuring prompt hazard detection. Experimental validation under standardized protocols confirmed continuous, synchronized data flow with no packet loss and stable updates under variable load conditions, within acceptable measurement error limits. The proposed architecture demonstrates robustness and reliability for data acquisition, monitoring, and decision support, effectively enhancing safety and operational efficiency in industrial environments.

Keywords: Industrial Safety Monitoring, Digital Twin Communication, Machine-Server Data Acquisition.