dc.contributor.author | Δαμικούκας, Σπυρίδων | el |
dc.contributor.author | Damikoukas, Spyridon | en |
dc.date.accessioned | 2023-09-25T09:56:04Z | |
dc.identifier.uri | https://dspace.lib.ntua.gr/xmlui/handle/123456789/58087 | |
dc.identifier.uri | http://dx.doi.org/10.26240/heal.ntua.25784 | |
dc.rights | Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα | * |
dc.rights.uri | http://creativecommons.org/licenses/by-nc-nd/3.0/gr/ | * |
dc.subject | Structural health monitoring | en |
dc.subject | Παρακολούθηση δομικής ακεραιότητας | el |
dc.subject | Seismic assessment | en |
dc.subject | Σεισμική αποτίμηση | el |
dc.subject | Accelerographs | en |
dc.subject | Επιταχυνσιογράφοι | el |
dc.subject | Ffast track pre seismic inspection method | en |
dc.subject | Μέθοδος ταχέως προσεισμικού ελέγχου | el |
dc.subject | Deep neural networks | en |
dc.subject | Βαθιά νευρωνικά δίκτυα | el |
dc.title | Structural Health Monitoring methods and Seismic Assessment | en |
dc.title | Μέθοδοι Παρακολούθησης Δομικής Ακεραιότητας και Σεισμική Αποτίμηση | el |
heal.type | doctoralThesis | |
heal.classification | Structural health monitoring | en |
heal.classification | Earthquake engineering | en |
heal.classification | Experimental engineering | en |
heal.classification | Ddeep neural networks | en |
heal.classification | Παρακολούθηση δομικής ακεραιότητας | el |
heal.classification | Αντισεισμική μηχανική | el |
heal.classification | Πειραματική μηχανική | el |
heal.classification | Βαθιά νευρωνικά δίκτυα | el |
heal.dateAvailable | 2024-09-24T21:00:00Z | |
heal.language | en | |
heal.access | free | |
heal.recordProvider | ntua | el |
heal.publicationDate | 2023-08-28 | |
heal.abstract | Buildings, structures in general, inevitably experience the effects of aging, as they are exposed to diverse environmental influences, external excitations, and the accumulation of stress and damage over time. To ensure their continued functionality and safety, regular maintenance is crucial. Structural Health Monitoring (SHM) plays a pivotal role in this endeavor. By implementing monitoring systems, engineers can continuously assess the structural integrity of buildings and identify any signs of deterioration or potential risks. In seismic regions, this becomes even more critical, as earthquakes can pose significant threats to infrastructure. SHM enables real-time data collection, allowing for early detection of seismic damages, and provides valuable insights for seismic assessment. By employing SHM technologies and conducting seismic assessments, authorities can proactively address vulnerabilities, minimize potential damage, and safeguard lives and properties during seismic events. The combination of aging structures and SHM underscores the importance of ongoing maintenance and proactive measures to ensure the resilience and longevity of our built environment in seismic-prone areas. In this scope, the focus of this dissertation in the first place was to create a monitoring system being as low cost as possible, but with maintaining the main principles of SHM: accuracy and reliability. The reason of choosing to build a low-cost system was to enable mass application, and thus generating new data, both structural and earthquake-performance ones. The monitoring system consists of a 20bit triaxial accelerometer, with sync capabilities for extensive Operational Modal Analysis (OMA) operation for deriving both structural eigenfrequencies and eigenmodes. Moreover, being versatile, it is a solution for permanent, semi-permanent or non-permanent monitoring, with internet connectivity in cases it is needed. The application of it as a monitoring system during the years of the current dissertation, counts over 41 structures, including typical buildings, bridges, heritage ones as also as museum exhibits. The next important section of the current dissertation, is the proposal of a methodology for fast track – first level – seismic assessment that includes a basic structural parameter for the dynamic response of structures, their fundamental eigenfrequency. By measuring it in ambient vibrations terms, it may not be the effective frequency engineers are waiting for in case of a strong seismic event, but it’s the elastic one, an upper “safe” limit, from which with appropriate modifications they are able to make safer assumptions for the current state of a structure. Without conflicting with any other known procedure in terms of a seismic assessment such as target displacement calculations or building taxonomy, the methodology refers to typical buildings that do satisfy the Single Degree of Freedom (SDOF) oscillator assumption, and by introducing an extra parameter in building stock / capacity models / fragility curves of the country to be applied, it introduces a new variability between buildings of the same category, adding an extra important experimental data and ability for differentiation. Moreover, thanks to the large number of monitored buildings, a sensitivity analysis was made between empirical code relationships regarding height and effective eigenperiod, and how they compare to experimental data after appropriate adjustments. Due to research interest in fascinating technologies such as Neural Networks, and specifically Deep Learning, and by seeing more and more applications that they do have a big impact two models were introduced in the context of research that was already be done. Regarding the monitoring system, a deep learning model is being introduced that its goal is to remove any additive noise from a “noisy” sensor that has recorded the ambient response of a building. As there were not any available labeled data, of simultaneously noisy and non-noisy measurements of the same building at the same time, artificial ones were generated by MDOF oscillators of different mass/stiffness and height combinationς. Thus, the model itself has the ability of getting a noisy acceleration timehistory in terms of image, denoise it and return an image, free of any additive noise, by preserving frequency spectrum and even revealing the fundamental eigenfrequency of a MDOF that previously was hidden under noise. The second model introduced, is a deep learning autoencoder that its goal is to predict the timehistory seismic response of a building by having as inputs its records under ambient vibrations and the earthquake itself. Again, in absence of data, artificially ones generated by various MDOF were used, and so the model by using images in spectrogram form came up to the challenges and predicted in good terms the response of any MDOF under 9 different earthquake scenarios. | en |
heal.sponsor | Scholarship of Research Committee of N.T.U.A. | en |
heal.sponsor | Υποτροφία ΕΛΚΕ ΕΜΠ | el |
heal.sponsor | EU-HORIZON 2020; ESPA2014-2020 | en |
heal.advisorName | Λαγαρός, Νίκος | el |
heal.advisorName | Lagaros, Nikos | en |
heal.committeeMemberName | Lagaros, Nikos | |
heal.committeeMemberName | Koziris, Nectarios | en |
heal.committeeMemberName | Mouzakis, Charalambos | en |
heal.committeeMemberName | Charmpis, Dimos | en |
heal.committeeMemberName | Vamvatsikos, Dimitrios | en |
heal.committeeMemberName | Sextos, Anastasios | en |
heal.committeeMemberName | Fragiadakis, Michalis | en |
heal.committeeMemberName | Λαγαρός, Νίκος | el |
heal.committeeMemberName | Κοζύρης, Νεκτάριος | el |
heal.committeeMemberName | Μουζάκης, Χαράλαμπος | el |
heal.committeeMemberName | Χαρμπής, Δήμος | el |
heal.committeeMemberName | Βαμβάτσικος, Δημήτριος | el |
heal.committeeMemberName | Σέξτος, Αναστάσιος | el |
heal.committeeMemberName | Φραγκιαδάκης, Μιχάλης | el |
heal.academicPublisher | Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Πολιτικών Μηχανικών. Τομέας Δομοστατικής. Εργαστήριο Στατικής και Αντισεισμικών Ερευνών | el |
heal.academicPublisherID | ntua | |
heal.numberOfPages | 267 σ. | el |
heal.fullTextAvailability | false |
The following license files are associated with this item: