Abstract:
This research investigates the feasibility of designing and developing Digital Twin
technology based on PLCnext technology, focusing on applications in automated
manufacturing. The research was divided into two phases, with Phase 1 emphasizing the study and testing of communication protocols between physical devices and digital models, and Phase 2 developing analytical and predictive capabilities using neural networks. The study found that the MQTT protocol demonstrates high efficiency in transmitting data from PLCnext devices to IoT platforms, while WebSocket is suitable for real-time data retrieval from platforms to Unity. The development of a hybrid CNN-LSTM neural network model provided more accurate manufacturing parameter predictions than single models, achieving mAP50
score of 63.8% compared to 57.2% for LSTM and 51.9% for CNN models. The factors most influencing process temperature were rotational speed (55%), torque (22%), and tool wear (15%). This study demonstrates that Digital Twin technology is an effective tool for improving manufacturing processes and predictive maintenance in Industry 4.0 environments.