Machine learning-driven nonlinear analysis of inclusion effects in aluminium alloys

Area of Science:

• Materials Science
• Mechanical Engineering
• Data Science

Background:

• Inclusions are critical defects in aluminum alloys, influencing material properties.
• Understanding the precise impact of inclusions on mechanical performance is vital for industrial applications.

Purpose of the Study:

• To comprehensively analyze the impact of inclusions on aluminum alloy properties using machine learning.
• To establish quantitative relationships between inclusion characteristics and material performance metrics.

Main Methods:

• Machine learning techniques, including SHAP value analysis, nonlinear regression, and cluster analysis.
• Comparative model evaluation using Random Forest and Gradient Boosting algorithms.
• Analysis of tensile strength, fatigue life, and corrosion behavior in relation to inclusion size.

Main Results:

• Inclusion size is the primary driver of mechanical property degradation, surpassing density's influence.
• Nonlinear regression identified critical inclusion size thresholds impacting tensile strength.
• Random Forest models demonstrated superior predictive accuracy for material properties.
• Significant improvements in strength (25 MPa) and fatigue life were observed with reduced inclusion size (5 μm vs. 10 μm).
• Corrosion rates exponentially increased with larger inclusion sizes.

Conclusions:

• Machine learning provides a robust framework for understanding inclusion-property relationships in aluminum alloys.
• Controlling inclusion size is crucial for enhancing material performance and reliability.
• The study offers actionable quality control parameters for aerospace and automotive industries.