A Transferable Digital Twin-Driven Process Design Framework for High-Performance Multi-Jet Polishing.

Area of Science:

• Manufacturing Engineering
• Materials Science
• Computer Science

Background:

• Multi-jet polishing (MJP) offers high accuracy for complex surfaces but faces challenges in efficiency, complexity, and stability, especially with changing conditions.
• Maintaining consistent polishing quality under variable working conditions remains a significant hurdle in MJP applications.

Purpose of the Study:

• To propose a digital twin (DT)-driven, human-centric design framework for optimizing multi-jet polishing (MJP).
• To enhance surface roughness prediction accuracy and robustness in MJP under varying operational parameters.
• To provide a predictive modeling solution for complex ultra-precision manufacturing under data scarcity and uncertainty.

Main Methods:

• Development of a digital twin (DT)-driven framework integrating key MJP factors: jet kinetic energy, nozzle structure, abrasive type, and machining path.
• Implementation of a feature-encoded transfer learning-based model for surface roughness prediction.
• Experimental validation using 3D printed workpieces under two distinct MJP working conditions.

Main Results:

• The proposed model demonstrated superior predictive performance compared to traditional methods.
• The model exhibited enhanced robustness and cross-condition adaptability in surface roughness prediction.
• Experimental results confirmed the effectiveness of the DT-driven framework and the transfer learning approach.

Conclusions:

• The developed predictive modeling component supports DT-driven process design for MJP.
• The framework offers a practical and extensible approach to optimize ultra-precision manufacturing processes.
• This work addresses challenges in data-scarce and uncertainty-dominated manufacturing environments.