A Bayesian physics-informed neural network model for probabilistic fatigue life assessment considering the size effect in additively manufactured materials

Key Points

• Fatigue life assessment reveals a probabilistic approach improves predictions for additively manufactured materials.
• The model effectively incorporates size effects, leading to more accurate results across different material dimensions.
• Employing a Bayesian neural network allows for better uncertainty quantification in fatigue life predictions.
• This approach highlights the need for advanced modeling techniques in understanding material performance under stress.