Inverse Design of Mechanical Metamaterials with Target Nonlinear Response via a Neural Accelerated Evolution...

Area of Science:

• Materials Science
• Mechanical Engineering
• Robotics

Background:

• Designing materials with specific nonlinear mechanical responses is crucial for advanced applications like soft robotics and energy absorption.
• Existing methods for achieving these responses are often complex and challenging to implement.

Purpose of the Study:

• To develop a versatile platform for realizing target nonlinear mechanical responses in materials.
• To establish an efficient computational method for designing materials with desired mechanical properties.

Main Methods:

• Utilized mechanical metamaterials based on hinged quadrilaterals, tuning their geometry to control internal rotations and mechanical response.
• Developed a neural network to establish a fast and inexpensive relationship between geometric parameters and stress-strain behavior.
• Integrated the neural network with an evolution strategy for efficient identification of optimal material geometries.

Main Results:

• Demonstrated that altering quadrilateral shapes effectively tunes internal rotations and achieves a wide spectrum of mechanical responses.
• The neural network accurately predicts stress-strain behavior based on geometric parameters with low computational cost.
• Successfully identified material geometries for optimized energy-absorbing systems, soft robots, and morphing structures.

Conclusions:

• Hinged quadrilateral mechanical metamaterials offer a flexible platform for achieving target nonlinear mechanical responses.
• The combination of neural networks and evolution strategies provides an efficient design pathway for advanced material applications.
• This approach facilitates the development of innovative soft robots, wearable devices, and energy-absorbing systems.