This thesis concerns the study, development and analysis of innovative artificial intelligence (AI)-driven optimization techniques within the System-by-Design (SbD) framework aimed at efficiently addressing the computational complexity inherent in advanced electromagnetic (EM) problems. By leveraging the available a-priori information as well as the proper integration of machine learning (ML) techniques with intelligent exploration strategies, the SbD paradigm enables the effective and reliable solution of the EM problem at hand, with user-defined performance and in a reasonable amount of time. The flexibility of the AI-driven SbD framework is demonstrated in practice with the implementation of two solution strategies to address the fully non-linear inverse scattering problem (ISP) for the detection and imaging of buried objects in ground penetrating radar (GPR)-based applications, and to address the design and optimization of mm-wave automotive radars that comply multiple challenging and contrasting requirements. A comprehensive set of numerical experiments is reported to demonstrate the efficacy and computational efficiency of the SbD-based optimization techniques in solving complex EM problems.

AI-Assisted Optimization Framework for Advanced EM Problems

Rosatti, Pietro
2024

Abstract

This thesis concerns the study, development and analysis of innovative artificial intelligence (AI)-driven optimization techniques within the System-by-Design (SbD) framework aimed at efficiently addressing the computational complexity inherent in advanced electromagnetic (EM) problems. By leveraging the available a-priori information as well as the proper integration of machine learning (ML) techniques with intelligent exploration strategies, the SbD paradigm enables the effective and reliable solution of the EM problem at hand, with user-defined performance and in a reasonable amount of time. The flexibility of the AI-driven SbD framework is demonstrated in practice with the implementation of two solution strategies to address the fully non-linear inverse scattering problem (ISP) for the detection and imaging of buried objects in ground penetrating radar (GPR)-based applications, and to address the design and optimization of mm-wave automotive radars that comply multiple challenging and contrasting requirements. A comprehensive set of numerical experiments is reported to demonstrate the efficacy and computational efficiency of the SbD-based optimization techniques in solving complex EM problems.
2-lug-2024
Inglese
Massa, Andrea
Salucci, Marco
Università degli studi di Trento
TRENTO
91
File in questo prodotto:
File Dimensione Formato  
PhD-Thesis.ROSATTI-Pietro.2024.pdf

embargo fino al 02/07/2026

Dimensione 3.65 MB
Formato Adobe PDF
3.65 MB Adobe PDF

I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.14242/179370
Il codice NBN di questa tesi è URN:NBN:IT:UNITN-179370