National Institute of the Republic of Serbia
ὅδε οἶκος, ὦ ἑταῖρε, μνημεῖον ἐστιν ζωῶν τῶν σοφῶν ἀνδρῶν, καὶ τῶν ἔργων αὐτῶν

PROGRAM

Plan rada Seminara Mehanika mašina i mehanizama - modeli i matematičke metode za MAJ 2023.

Ukoliko želite da učestvujete u radu seminara ili da postavite pitanja na kraju predavanja, a još niste registrovani na miteam platformi Matematičkog instituta, možete se registrovati popunjavanjem forme:
https://miteam.mi.sanu.ac.rs/asset/evJ49JBMGwQBc5FRz
Registrovani učesnici predavanja mogu pratiti na daljinu preko stranice:
https://miteam.mi.sanu.ac.rs/asset/cRwreFmETmbhEXnMG
Predavanja se mogu pratiti i bez registracije i učešća u diskusiji preko stranice:
https://miteam.mi.sanu.ac.rs/call/KzQPZsgH4eFSfz6Ap/0rIoev_XCjUed5FSoKE9HEh8aNVXqpfbA73m1eZiOZP
Arhiva snimljenih predavanja se nalazi na stranici:
https://miteam.mi.sanu.ac.rs/asset/PgqjStRApvcGxwtBx

UTORAK, 30.05.2023. u 17:00, sala 301f, Kneza Mihaila 36 i Live stream
Stepa M. Paunović, Mathematical Institute SANU, Belgrade, Serbia
DATA-DRIVEN MECHANICAL MODELLING – AWAITED SOLUTION OR A NEW PROBLEM?
The conventional approach to modelling mechanical behavior of materials relies on several observed universal laws, complemented by empirical constitutive laws formulated based on experiments and tests. These constitutive laws are usually phenomenological in nature, since they are derived from the observed macroscopic behavior of materials. Although these models are improved and enhanced by the fast developing field of micro- and nano-scale mechanics, they still represent idealization of the real world phenomena and introduce inherent and sometimes considerable approximation error, since they are founded on our incomplete knowledge of the natural world. On the other hand, in the last few years, rapid technological advance and enormous increase in computational power and available data have all led to a new modelling paradigm, where material behavior is modelled without any constitutive law whatsoever, but instead is predicted based solely on prior experience and data. This new modelling approach relies heavily on the use of neural networks and deep learning, and sees its opportunity for further development in the rise of the artificial intelligence. In this lecture we will examine and compare the basic characteristics of both modelling approaches, and try to conclude whether the new, data-driven modelling offers a long awaited solution, or represents a convenient and tempting, yet perhaps misleading path for mechanics and engineering.

Seminar Mehanika mašina i mehanizama - modeli i matematičke metode započeo je sa radom u junu 2018.god. Seminar se održava do dva puta mesečno, utorkom u periodu od 17.00 - 19.00 u Matematičkom institutu SANU.

dr Ivana Atanasovska
Rukovodilac seminara
Stepa Paunović
Sekretar seminara

﻿