Machine learning-driven design of optical transparency metamaterial absorbers with infrared-microwave compatible camouflage based on indium tin oxide

In this work, a feedforward neural network (FNN) and a direct inversion algorithm were employed to accelerate the design of an optically transparent metamaterial absorber (MMA) with infrared-microwave compatible camouflage properties. The MMA comprises an infrared reflection layer at the top, a microwave absorption layer in the middle, and a microwave reflection layer at the bottom. The designed MMA exhibits an infrared emissivity of approximately 0.3 within the 3–14 μm range and demonstrates microwave absorption rate exceeding 90% in the frequency range of 6.4–18 GHz. The microwave attenuation mechanisms of the MMA are analyzed using an equivalent circuit model, along with the distributions of the electric field, magnetic field, surface current, and power loss density. The measurement results of the fabricated sample exhibit strong consistency with the theoretical calculations and simulation results. This work provides novel insights into metamaterial design through the integration of neural networks and optimization algorithms, accelerating the development of multispectral compatible camouflage metamaterials.