Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Efficient Milling Quality Prediction with Explainable Machine Learning

Dennis Gross, Helge Spieker, Arnaud Gotlieb, Ricardo Knoblauch, Mohamed Elmansori

TL;DR

This work addresses predicting milling surface roughness with an explainable ML approach using random forest regression on a milling dataset for aluminum 2017A. By combining RF predictions with feature-importance analyses, it achieves accurate roughness predictions (MAPE < 8% for several metrics, and <5% for Rdqmaxmean) and identifies redundant sensors, particularly in force measurements. Removing non-significant sensors reduces hardware and maintenance costs while maintaining or improving predictive performance, illustrating the practical value of explainable ML in manufacturing. The study also examines robustness across data subsets and highlights the potential of integrating explainable models with digital twins for optimization, albeit noting overfitting risks from limited data and the need for broader validation.

Abstract

This paper presents an explainable machine learning (ML) approach for predicting surface roughness in milling. Utilizing a dataset from milling aluminum alloy 2017A, the study employs random forest regression models and feature importance techniques. The key contributions include developing ML models that accurately predict various roughness values and identifying redundant sensors, particularly those for measuring normal cutting force. Our experiments show that removing certain sensors can reduce costs without sacrificing predictive accuracy, highlighting the potential of explainable machine learning to improve cost-effectiveness in machining.

Efficient Milling Quality Prediction with Explainable Machine Learning

TL;DR

This work addresses predicting milling surface roughness with an explainable ML approach using random forest regression on a milling dataset for aluminum 2017A. By combining RF predictions with feature-importance analyses, it achieves accurate roughness predictions (MAPE < 8% for several metrics, and <5% for Rdqmaxmean) and identifies redundant sensors, particularly in force measurements. Removing non-significant sensors reduces hardware and maintenance costs while maintaining or improving predictive performance, illustrating the practical value of explainable ML in manufacturing. The study also examines robustness across data subsets and highlights the potential of integrating explainable models with digital twins for optimization, albeit noting overfitting risks from limited data and the need for broader validation.

Abstract

This paper presents an explainable machine learning (ML) approach for predicting surface roughness in milling. Utilizing a dataset from milling aluminum alloy 2017A, the study employs random forest regression models and feature importance techniques. The key contributions include developing ML models that accurately predict various roughness values and identifying redundant sensors, particularly those for measuring normal cutting force. Our experiments show that removing certain sensors can reduce costs without sacrificing predictive accuracy, highlighting the potential of explainable machine learning to improve cost-effectiveness in machining.
Paper Structure (11 sections, 2 equations, 5 figures)

This paper contains 11 sections, 2 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodolodgy
  4. Random Forest Regression Models
  5. Explainable ML Methods
  6. Case Study
  7. ML Model Training and Evaluation
  8. Assessing Gini Importance in Random Forest Regression Models to reduce number of sensors
  9. Assessing the Significance of Feature Permutation Importance Across Various Data Subsets
  10. Discussion
  11. Conclusion

Figures (5)

  • Figure 1: Milling machine that produces workpieces.
  • Figure 2: The ML prediction model receives the box plots (for time and frequency domains) and machine configuration parameters to output the quality measure (surface roughness).
  • Figure 3: Predictive model test results with mean squared error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). .
  • Figure 4: Feature Importance rankings across the different roughness prediction random forests (see labels). The feature ranking is identical for all prediction targets.
  • Figure 5: Feature Permutation Importance across various data subsets.