An Improved Entropy Regularization Method for Defect Detection in Multilayer Composites

Abstract

Microwave time reversal (MTR) as a microwave imaging method can provide autofocus and super-resolution with real-time, efficient, high-contrast, and low-cost properties. It is applied here for nondestructive testing (NDT) of small defects buried in multilayer composites backed by metal. It is critical to estimate the focusing time with localization methods in MTR. The traditional entropy regularization method (ERM) may give the wrong focusing time estimation for the strong reflection from background metal. This leads to a wrong localization of the defect. An improved ERM (IERM) is proposed here by properly combining ERM and target initial reflection method (TIRM) to ensure a correct localization. MTR based on IERM (IERM-MTR) is then applied to detect the defect in multilayer composites when the permittivities of each layer are known. Its stability, accuracy, and noise tolerance are proved by multiple numerical investigations. An iterative strategy based on the IERM-MTR and particle swarm optimization (PSO) method is also proposed here for detecting the defect in multilayer composites when the permittivities of each layer are unknown. Numerical investigations are carried out to prove its effectiveness. Small defects are easily detected in multilayer composites applied here.