Digital twin system for manufacturing processes based on a multi-layer knowledge graph model.

Area of Science:

• Manufacturing Engineering
• Computer Science
• Data Science

Background:

• Digital twin technology in manufacturing faces challenges integrating diverse data sources and managing real-time data flow.
• Effective digital twin modeling requires robust data integration and processing capabilities.

Purpose of the Study:

• To propose and validate a novel three-layer knowledge graph architecture for enhancing digital twin modeling in manufacturing processes.
• To address challenges in data integration and real-time data flow for digital twins.

Main Methods:

• Developed a three-layer knowledge graph architecture: concept layer (knowledge network), model layer (digital-physical alignment), and decision layer (data-driven decision support).
• Integrated multi-source data and validated the system in aero-engine blade production.
• Utilized the architecture for predictive analysis, anomaly detection, process control, and quality management.

Main Results:

• The system successfully integrated multi-source data, enhancing predictive analysis and anomaly detection.
• Maximum contour error precision improved from 0.073 mm to 0.062 mm over a 5-month validation period.
• Product qualification rate increased from 81.3% to 85.2%.

Conclusions:

• The proposed knowledge graph architecture significantly enhances digital twin utilization in manufacturing.
• The system demonstrates robust capabilities for process control, quality management, and predictive maintenance.
• This approach offers a scalable solution for advancing digital twin technology in complex manufacturing environments.