A deep learning dataset for metal multiaxial fatigue life prediction

Area of Science:

• Materials Science
• Mechanical Engineering
• Artificial Intelligence

Background:

• Multiaxial fatigue failure in metals causes significant industrial losses.
• Deep learning shows promise for predicting metal fatigue life, but requires extensive training data.
• Current data collection for fatigue testing is costly and labor-intensive.

Purpose of the Study:

• To create a comprehensive, high-quality dataset for multiaxial fatigue life prediction.
• To alleviate the scarcity of training data for deep learning models in fatigue analysis.
• To facilitate research at the intersection of artificial intelligence and metal fatigue.

Main Methods:

• Compiled a dataset of 1167 samples from 40 different materials sourced from existing literature.
• Included critical mechanical properties such as elastic modulus, yield strength, tensile strength, and Poisson's ratio.
• Incorporated 48 distinct loading paths and additional data like composition ratios and processing conditions.

Main Results:

• The created dataset was validated using common deep learning models, demonstrating its effectiveness.
• The dataset provides a robust foundation for training and testing AI models for fatigue life prediction.
• Successfully addressed the challenge of data scarcity in the field of metal fatigue research.

Conclusions:

• The developed dataset is a valuable resource for researchers in AI and metal fatigue.
• It enables more accurate and efficient deep learning-based predictions of multiaxial fatigue life.
• This work advances the application of AI in materials science for predicting material failure.