Ultra-Broadband Microwave Absorption and Programmable Multispectral Camouflage Enabled by Neural-Network-Driven Impedance-Gradient Metadevices

Abstract Achieving omnidirectional multispectral compatible camouflage remains a significant challenge due to the pronounced disparities in electromagnetic wavelengths and constraints on the response mechanisms of natural materials. Herein, neural networks are employed to intelligently optimize the design of multiscale impedance-gradient (IG) metadevices tailored for multispectral compatibility. The macro-gradient unit is engineered with precise impedance matching and high rotational symmetry to provide exceptional microwave ultra-broadband absorption (2–18 GHz), with insensitivity angles reaching 60°. By integrating a polyimide foam substrate with an MXene-functionalized nanostructured photochromic top layer, the finalized device exhibits remarkable infrared thermal insulation (ΔT $$\approx$$ ≈ 65 °C) and low emissivity (0.38), alongside rapid visible color change (1 ~ 2 s) enabled by nanoscale photochromic switching. Furthermore, IG metadevices deliver programmability and multimodality, alongside impact resistance (~ 30,000 N) and environmental stability. This work provides novel paradigms for the intelligent design of multispectral compatible camouflage systems adaptable to complex scenarios.