Today’s manufacturing systems are getting more complex because there is a strongneedto use energy in a better way and to make smart and clear decisions. This thesis presentsa new framework called the Causal Digital Twin Framework (CDTF), whichhelpstomake better and more sustainable decisions in smart factories. Unlike the previouspublished Digital Twin frameworks, that mostly do simulations and monitoring, CDTFworks with real-time data, connects physical and digital parts, uses AI tomakepredictions, and finds the real causes behind problems. This framework has four main layers: Data acquisition and Integration layer, Cyber-Physical Synchronization Layer, AI-Driven Decision-Making Layer, andCausalInference and Modeling Layer. Layer 1 is used to collect real time data fromthephysical system. Layer 2 connects the physical and virtual parts, making surethedatamoves and works smoothly between them. In Layer 3, an XGBoost model was usedtopredict how long the manufacturing system parts will keep working, andit gaveaccurate results. Layer 4 moved from just finding connections to understandingrealcauses, using specific questions, causal models, OLS regression, and “what if”testswith the DoWhy tool. It was tested on a real 3D printer (Prusa i3 MK3S), where more than 9.6 millionsensordata points were collected using MQTT, handled with Node-RED, and savedinaTimescaleDB system. Also, live dashboards were made with Grafana toshowpredictions, problems, and cause-related insights in real time. The results show that the CDTF helps with both predicting when maintenance is neededand finding the real reasons behind energy use and system problems. By usingAI andcausal analysis in a clear and well-structured way, this research improvestheapplication of the Digital Twin technology to make it more transparent, understandable,and focused on saving energy. This new framework also supports buildingsmartsystems that can work on their own, can be repeated in other places, and followethicaland responsible decision-making.
- Innovation and Product Management
A Causal Digital Twin Framework for Sustainable Decision-Making in Smart Manufacturing Systems
Menghwar, A. K. (Author). 2025
Student thesis: Master's Thesis