Synergistic electromagnetic mechanisms and intelligent inverse design of ultra-wideband carbon/silicon carbide fiber metamaterials

To address the bottleneck of multidimensional parameter space optimization faced by traditional absorbers in broadband design, this study proposes an intelligent inverse design framework integrating a General Regression Neural Network optimized by the Sparrow Search Algorithm with a segmented particle swarm algorithm. By configuring an impedance gradient through an upper sparse H-shaped array and a lower high-density lantern-shaped array, efficient energy dissipation is achieved by combining the magnetic coupling resonance of silicon carbide fibers with the strong eddy current loss on carbon fiber surfaces. Both experimental measurements and simulation results confirm that the 3-mm-thick anisotropic structure-comprising H-shaped and lantern-shaped elements-exhibits a wide effective absorption bandwidth of 10.2 GHz and a minimum reflection loss of -19.32 dB. However, its absorption performance in the low-frequency band remains limited. In contrast, the double-layer co-directional arrangement structure, by optimizing synergistic electromagnetic coupling and impedance matching between heterogeneous units, more effectively excites cooperative electromagnetic loss between units. This achieves an effective absorption bandwidth of 14.0675 GHz across the 3.9325–18 GHz frequency band while maintaining a minimum reflection loss of -19.80 dB. Simultaneously, radar cross-section testing validates its significant scattering suppression capability across a wide angular domain, demonstrating intelligent design for high-performance stealth materials.