Machine learning-enabled inverse design of bioinspired layered composite structures with maximum auxetic performance.

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Design

Background:

• Biological tissues inspire layered composites with auxetic properties (negative Poisson's ratio).
• Optimizing auxetic performance in high-dimensional laminate designs is complex.
• Identifying specific layups for desired auxetic responses requires advanced methods.

Purpose of the Study:

• To introduce an inverse design framework for identifying laminate layups with minimum Poisson's ratio.
• To efficiently search high-dimensional design spaces for auxetic composite structures.
• To provide a physics-grounded and data-efficient route for engineering auxetic materials.

Main Methods:

• Combines multi-start resampling with machine learning-guided clustering.
• Utilizes laminate mechanics to link ply angles to effective properties.
• Validates predictions through computer simulations and laboratory measurements.

Main Results:

• Resolves three distinct layup categories driving auxetic expansion.
• Explains the role of shear-strain mismatch in through-thickness auxetic behavior.
• Identifies layups approaching lower Poisson's ratios under practical constraints.

Conclusions:

• The framework offers a physics-grounded, data-efficient method for auxetic composite design.
• Provides concise design rules for applications like impact mitigation and vibration control.
• Successfully reproduces known minima and discovers new auxetic layups.