A Regularized Regression Thermal Error Modeling Method for CNC Machine Tools under Different Ambient Temperatures and...

Area of Science:

• Manufacturing Engineering
• Metrology
• Machine Learning

Background:

• Thermal errors significantly impact CNC machine tool precision.
• Existing deep learning models for thermal error prediction are complex and data-intensive.
• Interpretability and practical implementation remain challenges in current methods.

Purpose of the Study:

• To propose a novel, interpretable, and practical regularized regression algorithm for CNC machine tool thermal error modeling.
• To enable automatic selection of temperature-sensitive variables for improved model efficiency.
• To demonstrate superior performance compared to existing state-of-the-art algorithms.

Main Methods:

• Utilized least absolute regression combined with two regularization techniques.
• Developed a simplified model structure for ease of implementation and interpretability.
• Implemented automatic temperature-sensitive variable selection.
• Compared prediction accuracy and robustness against deep learning algorithms.

Main Results:

• The proposed regularized regression model achieved the highest prediction accuracy.
• The method demonstrated superior robustness compared to existing algorithms.
• Automatic variable selection enhanced model efficiency.
• Compensation experiments confirmed the model's practical effectiveness.

Conclusions:

• The proposed regularized regression algorithm offers an effective and interpretable solution for CNC machine tool thermal error modeling.
• This approach provides a practical alternative to complex deep learning methods, requiring less data and offering better insights.
• The validated effectiveness paves the way for improved precision in CNC machining operations.