Machine Learning-Based Prediction of Specific Energy Consumption for Cut-Off Grinding

Area of Science:

• Manufacturing Engineering
• Materials Science
• Data Science

Background:

• Cut-off operations are crucial in manufacturing but are highly energy-intensive.
• Predicting specific energy consumption (SEC) is key to understanding and reducing energy use in grinding processes.

Purpose of the Study:

• To develop and validate a novel methodology for predicting SEC in oxygen-free copper (OFC-C10100) cut-off grinding.
• To employ supervised machine learning techniques for accurate energy consumption prediction.

Main Methods:

• Designed a state-of-the-art experimental setup for abrasive cutting of OFC-C10100.
• Utilized Gaussian process regression, regression trees, and artificial neural network (ANN) to predict energy consumption based on feed rate, cutting thickness, and tool type.
• Evaluated SEC using predicted energy consumption values.

Main Results:

• Gaussian process regression demonstrated superior performance with minimal validation and testing errors.
• Predicted energy consumption values accurately evaluated SEC, achieving a correlation coefficient of 0.98.
• The relationship between predicted SEC and material removal rate aligned with physical models, confirming prediction accuracy.

Conclusions:

• Supervised machine learning, particularly Gaussian process regression, offers a reliable method for predicting specific energy consumption in cut-off grinding.
• The developed methodology provides an accurate tool for analyzing and optimizing energy efficiency in the manufacturing industry.