Implementation of Cyber-Physical Production Systems for Quality Prediction and Operation Control in Metal Casting

Area of Science:

• Manufacturing Engineering
• Industrial Informatics
• Artificial Intelligence in Manufacturing

Background:

• Early prediction of internal defects in metal casting is crucial for reducing costs and optimizing production schedules.
• Cyber-physical Production Systems (CPPS) offer a viable solution for real-time quality prediction and operational control.

Purpose of the Study:

• To design and implement a CPPS framework for quality prediction and operation control in a real metal casting factory.
• To demonstrate the practical application of CPPS integrating various technologies for enhanced manufacturing processes.

Main Methods:

• Designed a CPPS architecture integrating IoT, AI, simulations, MES, and APS.
• Deployed temperature sensors and IoT devices on casting machines for real-time data collection.
• Utilized HBase for IoT data repository and Spark for data pre-processing.
• Applied and compared machine learning algorithms (Decision Tree, Random Forest, ANN, SVM) for quality prediction.

Main Results:

• Successfully implemented a CPPS framework in an actual factory setting.
• Demonstrated the system's operation through a CPPS dashboard, showcasing quality prediction and control.
• Quantitatively verified the usefulness and benefits of the implemented CPPS.

Conclusions:

• The study provides a specific architectural framework, use cases, software, and methodologies for CPPS implementation in metal casting.
• The successful verification of CPPS's quantitative effects is expected to promote its adoption in the manufacturing industry.