Application of neural network algorithms in the design of electromagnetic parameters for dielectric loss microwave absorbing materials

Dielectric loss microwave absorbing materials have been widely used to mitigate electromagnetic interference and achieve radar stealth. Their design and optimization involve multiple electromagnetic parameters, such as dielectric constant, magnetic permeability, and loss factors. Traditional optimization methods typically require extensive computation and experimentation, making them inefficient and prone to local optima. In recent years, neural network algorithms, as effective nonlinear modeling and optimization tools, have found growing applications in optimizing the performance of absorbing materials. This paper explores the application of neural networks in optimizing electromagnetic parameters of dielectric loss microwave absorbing materials. Using neural networks, we predict the reflectivity values under different electromagnetic parameters and examine the patterns of variation with frequency and thickness. The results show that optimal reflectivity for various frequencies corresponds to different sets of electromagnetic parameters and matching thicknesses. As the frequency increases, the optimal dielectric constant values (both real and imaginary parts) concentrate in a lower range, especially when the material thickness is small. Finally, the consistency between predicted and experimentally measured reflectivity values confirms the reliability of the neural network-based predictions.