Knowledge extraction and transfer in data-driven fracture mechanics

Area of Science:

• Fracture Mechanics
• Data-Driven Science
• Computational Materials Science

Background:

• Data scarcity hinders knowledge extraction and transfer in fracture mechanics.
• Current data-driven methods are limited in addressing these challenges.
• Effective knowledge extraction and transfer are crucial for advancing the field.

Purpose of the Study:

• To propose and demonstrate a data-driven framework for knowledge extraction in fracture mechanics.
• To address the challenge of data scarcity in knowledge extraction.
• To explore the potential of knowledge transfer among different fracture problems.

Main Methods:

• Developed a data-driven framework for knowledge extraction with accuracy assessment metrics.
• Applied a tailored active learning method for knowledge extraction in a data-limited regime.
• Demonstrated knowledge transfer by connecting 3D and 2D fracture mechanics problems.

Main Results:

• Accurate knowledge extraction was achieved even with limited data using active learning.
• The framework successfully enabled knowledge transfer between distinct fracture mechanics problems.
• The proposed methods proved effective for a small-scale toughness measurement problem.

Conclusions:

• The data-driven framework offers a viable solution for knowledge extraction and transfer in fracture mechanics.
• Active learning is effective in overcoming data scarcity for knowledge extraction.
• Knowledge transfer holds significant potential for transformative impact in the field.