Weld Surface Geometry's Impact on Generalizability of Machine Learning Models for Fatigue Life Prediction

Dies ist die aktuellste Version dieses Eintrags.

Kurzfassung

Designing welded steel structures for fatigue resistance under cyclic loading requires accurate assessment of local weld geometry effects, which traditional methods often oversimplify. This study employs Machine Learning (ML) to predict fatigue life based on XGBoost models pretrained on flux cord arc welded (FCAW) and submerged arc welded (SAW) butt welds. Extending the dataset to include Laser Hybrid (LH) welds and specimens with weld defects, the research explores ML model transferability across different welding methods and defect types defined by DIN standards. Results highlight challenges in predicting fatigue life accurately when weld surface geometry varies or significant defects are present, emphasizing the need for robust anomaly detection in 3D laser scanning data. The study underscores the nuanced impact of locally occurring weld defects and imperfections on fatigue life predictions, revealing limitations in current ML.

Verfügbare Versionen dieses Eintrags

• Weld Surface Geometry's Impact on Generalizability of Machine Learning Models for Fatigue Life Prediction. (deposited 13 Jan 2025 07:52) [Gegenwärtig angezeigt]