Developing a carpet cloak operating for a wide range of incident angles using a deep neural network and PSO...

Area of Science:

• Electromagnetics and Optics
• Materials Science
• Computational Physics

Background:

• Metasurface-based carpet cloaks offer reduced loss and easier fabrication compared to traditional cloaking methods.
• A key limitation of current carpet cloaks is their performance dependency on the incident wave angle.
• Advanced design strategies are needed to achieve broadband and wide-angle invisibility cloaking.

Purpose of the Study:

• To design a metasurface-based carpet cloak with wide-angle operational capabilities.
• To leverage computational optimization and machine learning for accelerated cloak design.
• To numerically validate the cloaking performance across a significant range of incident angles.

Main Methods:

• Utilized the Particle Swarm Optimization (PSO) algorithm for metasurface unit cell design.
• Integrated a trained deep neural network to expedite the calculation of reflection phases.
• Performed numerical analysis of the designed carpet cloak's near-field and far-field responses.

Main Results:

• Successfully developed a metasurface carpet cloak operational for incident angles from 0 to 65 degrees.
• The neural network significantly accelerated the design and optimization process.
• Numerical simulations confirmed excellent cloaking performance across the specified wide angle range.

Conclusions:

• The proposed PSO and neural network-assisted design approach enables the creation of wide-angle metasurface carpet cloaks.
• This method overcomes the angular limitations of previous carpet cloak designs.
• The developed cloak demonstrates significant potential for practical invisibility applications.