This thesis examines the integration of sustainable nanomaterials and nanostructures into Micro- and Nano-Electro-Mechanical systems (MEMS/NEMS), with a focus on developing nonlocal continuum formulations for small-scale structures. Fundamental concepts of nonlocal elasticity are introduced, including Eringen’s strain-driven theory, the role of averaging kernels in constitutive laws, and the resolution of paradoxes through stress-driven formulations. A displacementdriven nonlocal elastic model is exploited to analyze the dynamics of nanobeams on nanofoundations, including case studies on nanocantilevers and functionally graded nanobeams under several boundary conditions, revealing size-dependent stiffening effects. To overcome challenges in solving complex nonlocal problems, Physics-Informed Neural Networks (PINNs) are proposed as a powerful computational tool, demonstrating their efficacy in eigenvalue problems for nonlocal beams. The methodology is extended to two-dimensional continua using stress-driven and displacement-driven nonlocal approaches, evaluating structural responses under distributed loadings for nanoplates with clamped and simply supported edges. The presented approaches provide accurate solutions for static and dynamic analyses, offering insights into the design of sustainable nanodevices for applications in structural health monitoring and electromechanical systems.
Questa tesi esamina l’integrazione di nanomateriali e nanostrutture sostenibili nei sistemi Micro- e Nano-Elettro-Meccanici (MEMS/NEMS), con particolare attenzione allo sviluppo di formulazioni continue non locali per strutture di piccola scala. Vengono introdotti i concetti fondamentali dell’elasticità non locale, inclusi la teoria strain-driven di Eringen, il ruolo dei kernel di mediazione nelle leggi costitutive e la risoluzione dei paradossi mediante formulazioni stress-driven. Un modello elastico non locale displacement-driven viene impiegato per analizzare la dinamica di nanotravi su nanofondazioni, includendo casi studio relativi a nanocantilever e nanotravi a gradazione funzionale soggette a diverse condizioni al contorno, evidenziando effetti di irrigidimento dipendenti dalla scala dimensionale. Per superare le difficoltà legate alla risoluzione di complessi problemi non locali, le Physics-Informed Neural Networks (PINNs) vengono proposte come potente strumento computazionale, dimostrando la loro efficacia nei problemi agli autovalori per travi non locali. La metodologia viene inoltre estesa a continui bidimensionali mediante approcci non locali stress-driven e displacement-driven, valutando le risposte strutturali sotto carichi distribuiti per nanopiastre con bordi incastrati e semplicemente appoggiati. Gli approcci presentati forniscono soluzioni accurate per analisi statiche e dinamiche, offrendo nuove prospettive per la progettazione di nanodispositivi sostenibili destinati ad applicazioni nel monitoraggio della salute strutturale e nei sistemi elettromeccanici.
Physics-informed neural networks (PINNs) in nonlocal elasticity [Reti neurali Informate dalla fisica (PINNs) nell’elasticità non locale] / Das, B.. - (2026 Jun 25).
Physics-informed neural networks (PINNs) in nonlocal elasticity [Reti neurali Informate dalla fisica (PINNs) nell’elasticità non locale]
DAS, BAIDHEI
2026-06-25
Abstract
This thesis examines the integration of sustainable nanomaterials and nanostructures into Micro- and Nano-Electro-Mechanical systems (MEMS/NEMS), with a focus on developing nonlocal continuum formulations for small-scale structures. Fundamental concepts of nonlocal elasticity are introduced, including Eringen’s strain-driven theory, the role of averaging kernels in constitutive laws, and the resolution of paradoxes through stress-driven formulations. A displacementdriven nonlocal elastic model is exploited to analyze the dynamics of nanobeams on nanofoundations, including case studies on nanocantilevers and functionally graded nanobeams under several boundary conditions, revealing size-dependent stiffening effects. To overcome challenges in solving complex nonlocal problems, Physics-Informed Neural Networks (PINNs) are proposed as a powerful computational tool, demonstrating their efficacy in eigenvalue problems for nonlocal beams. The methodology is extended to two-dimensional continua using stress-driven and displacement-driven nonlocal approaches, evaluating structural responses under distributed loadings for nanoplates with clamped and simply supported edges. The presented approaches provide accurate solutions for static and dynamic analyses, offering insights into the design of sustainable nanodevices for applications in structural health monitoring and electromechanical systems.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi_Baidehi_per_stampa.pdf
embargo fino al 25/06/2027
Tipologia:
Tesi di dottorato
Licenza:
NON PUBBLICO - Accesso privato/ristretto
Dimensione
10.31 MB
Formato
Adobe PDF
|
10.31 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


