The fully non-linear inverse scattering problem (ISP) is addressed as a global optimization problem and solved with high computational efficiency. Towards this end, a novel system-by-design (SbD) approach is proposed leveraging on the suitable interconnection of functional blocks aimed at (i) 'smartly' defining a limited-dimensionality yet highly-flexible set of degrees-of-freedom (DoFs), (ii) formulating a suitable cost function to minimize, and (iii) integrating a global optimization approach based on evolutionary algorithms (EAs) with a computationally-fast digital twin (DT) of the accurate (but time-consuming) full-wave solver, which is adaptively 'reinforced' while guiding the optimization towards the global optimum. An illustrative example is shown to assess the effectiveness and the high computational efficiency of the proposed inversion method.
AI-Assisted Computationally-Efficient Global Optimization for Inverse Scattering
Hannan M. A.;
2021-01-01
Abstract
The fully non-linear inverse scattering problem (ISP) is addressed as a global optimization problem and solved with high computational efficiency. Towards this end, a novel system-by-design (SbD) approach is proposed leveraging on the suitable interconnection of functional blocks aimed at (i) 'smartly' defining a limited-dimensionality yet highly-flexible set of degrees-of-freedom (DoFs), (ii) formulating a suitable cost function to minimize, and (iii) integrating a global optimization approach based on evolutionary algorithms (EAs) with a computationally-fast digital twin (DT) of the accurate (but time-consuming) full-wave solver, which is adaptively 'reinforced' while guiding the optimization towards the global optimum. An illustrative example is shown to assess the effectiveness and the high computational efficiency of the proposed inversion method.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
