Stochastic Multiscale Analysis; Bayesian Multiscale Update

Ιμπραημάκης, Μάριος; Impraimakis, Marios

dc.contributor.author	Ιμπραημάκης, Μάριος	el
dc.contributor.author	Impraimakis, Marios	en
dc.date.accessioned	2018-05-03T10:27:07Z
dc.date.available	2018-05-03T10:27:07Z
dc.date.issued	2018-05-03
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/46922
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.8172
dc.description	Εθνικό Μετσόβιο Πολυτεχνείο--Μεταπτυχιακή Εργασία. Διεπιστημονικό-Διατμηματικό Πρόγραμμα Μεταπτυχιακών Σπουδών (Δ.Π.Μ.Σ.) “Δομοστατικός Σχεδιασμός και Ανάλυση των Κατασκευών”	el
dc.rights	Default License
dc.subject	Bayesian Update	en
dc.subject	Subset Simulation	en
dc.subject	Multiscale Modeling	en
dc.subject	Hierarchical Strategy	en
dc.subject	Nanocomposites	en
dc.subject	Μπεϋζιανή Ενημέρωση	el
dc.subject	Μέθοδος Υποσυνόλων	el
dc.subject	Ανάλυση πολλαπλών κλιμάκων	el
dc.subject	Ιεραρχική Μοντελοποίηση	el
dc.subject	Νανοσύνθετα Υλικά	el
dc.title	Stochastic Multiscale Analysis; Bayesian Multiscale Update	en
heal.type	masterThesis
heal.classification	Computational Stochastic Mechanics	en
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2018-03-07
heal.abstract	Bayesian updating is a powerful method to learn and calibrate models with data and observations, facts that is of utmost importance in multiscale problems with uncertain microscale status like very random and hard predicted nanocomposite behavior. In this work BUS (Bayesian Updating with Structural reliability methods) with SuS (Subset Simulation) in a multiscale environment is employed to compute the posterior distribution of microscale random parameters in a framework that microscale with mesoscale and microscale with macroscale pair models converge into each experimental data simultaneously. More specific, every sample cluster of every subset within SuS in this parallel double Bayesian problem is forced to agree with the other one. In the end, the samples in the final subset (posterior samples in Bayesian terms) have the best agreement with experimental data. This methodology is very promising for nanomaterial reinforced composites which have big uncertainty range with quite unexpected measurements and really large number of parameters. It is a gainful direction for engineering practice and non-costly experimental investigations, being concurrently quite appropriate for every multiscale modeling application.	en
heal.sponsor	Supported by the Bodossaki Foundation	en
heal.advisorName	Παπαδόπουλος, Βησσαρίων	el
heal.committeeMemberName	Κουμούσης, Βλάσης	el
heal.committeeMemberName	Φραγκιαδάκης, Μιχαήλ	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Πολιτικών Μηχανικών	el
heal.academicPublisherID	ntua
heal.numberOfPages	61 σ.	el
heal.fullTextAvailability	true