Reconfigurable model clusters for scalable modelling of feed drive dynamics

Area of Science:

• Mechanical Engineering
• Control Systems Engineering
• Computational Modeling

Background:

• Accurate modeling of machine tool feed drives is crucial for performance optimization and predictive maintenance.
• Traditional methods require extensive individual machine calibration, hindering scalability.
• The increasing demand for digital twins and shadows necessitates more efficient modeling techniques.

Purpose of the Study:

• To develop a novel, scalable approach for modeling machine tool feed drive dynamics.
• To reduce the need for individual machine calibration in developing digital twins and shadows.
• To leverage Bayesian model selection for accurate representation of machine behavior.

Main Methods:

• Utilized a cluster of pre-calibrated models for a fleet of similar machines or varying conditions.
• Applied Bayesian model selection to assimilate internal controller signals into the model cluster.
• Selected an optimal combination of models to represent individual machines accurately.

Main Results:

• Demonstrated the effectiveness of the approach through numerical simulations with known ground truths.
• Showcased the potential to simplify and scale feed drive modeling significantly.
• Identified key technical and operational considerations for practical implementation.

Conclusions:

• The proposed approach offers a simplified and scalable solution for feed drive modeling in machine tools.
• Facilitates large-scale development of machine tool digital twins and shadows without extensive calibration.
• Provides a foundation for broader application by outlining necessary technical and operational considerations.