Enhanced Electromagnetic Imaging of Imperfectly Conducting Cylinders Using Hybrid Genetic Algorithm and Deep Learning

Abstract

:Electromagnetic imaging (EMI) plays a crucial role in various applications, including medical 
diagnostics, geophysical exploration, and non-destructive testing. However, accurately reconstructing the 
properties of imperfectly conducting cylinders remains a significant challenge due to the ill-posed nature 
of inverse scattering problems. This study presents an enhanced electromagnetic imaging approach that 
integrates a hybrid Genetic Algorithm (GA) with Deep Learning (DL) techniques to improve reconstruction 
accuracy and computational efficiency. The proposed method utilizes GA for global optimization, ensuring 
robust parameter estimation, while a deep neural network refines the inversion process by learning complex 
scattering patterns. Numerical simulations and experimental results demonstrate that the hybrid approach 
outperforms traditional inversion methods in terms of accuracy, convergence speed, and resilience to noise. 
The findings indicate that combining evolutionary algorithms with deep learning offers a powerful 
framework for solving complex inverse scattering problems, making it a promising tool for advanced 
electromagnetic imaging applications.