An AI-Driven Method for Microwave NDT/NDE within the System-by-Design Framework

This work presents an innovative microwave imaging (MI) method for non-destructive testing and evaluation (NDT/NDE). By leveraging the System-by-Design (SbD) paradigm, the proposed technique uses a differential formulation to exploit a-priori knowledge of the unperturbed structure under test (SUT) and on the expected anomalies within it. A customized global optimization strategy is designed to solve the underlying inverse scattering problem (ISP), while mitigating its computational complexity by means of properly-implemented and interconnected "functional blocks". These blocks are designed to define a "smart" minimum-cardinality solution descriptor set and create a fast but accurate digital twin (DT) of the forward scattering solver. An illustrative numerical example, concerned with the inspection of water-infiltrated concrete, is shown to demonstrate the effectiveness of the method.