A machine learning framework predicts fatigue life in additively manufactured stainless steel, indicating key factors like elongation and applied stress.
Key Points
The aim is to develop a machine learning framework for predicting the fatigue life of additively manufactured 316L stainless steel based on mechanical properties and loading conditions.
Proposed a machine learning framework for fatigue life prediction without requiring detailed AM process data.
Utilized a literature-derived dataset covering multiple AM techniques for model training and comparison.
Employed multilayer perceptron for predictive performance and Explainable AI for variable interpretation.
Identified applied stress, elongation, and strength-related properties as dominant factors in fatigue life.
Demonstrated feasibility of property-based fatigue life prediction across various additive manufacturing processes.