Προσομοίωση της καταλυτικής πυρόλυσης με χρήση μηχανικής μάθησης και ενσωμάτωσης δεδομένων

Μπλήτας, Δημήτριος; Blitas, Dimitrios

dc.contributor.author	Μπλήτας, Δημήτριος	el
dc.contributor.author	Blitas, Dimitrios	en
dc.date.accessioned	2023-06-09T07:43:01Z
dc.date.available	2023-06-09T07:43:01Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/57809
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.25506
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	Μηχανική μάθηση	el
dc.subject	Καταλυτική Πυρόλυση	el
dc.subject	Νευρωνικά δίκτυα	el
dc.subject	Ενσωμάτωση δεδομένων	el
dc.subject	Βιομηχανική προσομοίωση	el
dc.subject	Machine learning	en
dc.subject	Catalytic cracking	en
dc.subject	Neural networks	en
dc.subject	Data embedding	en
dc.subject	Industrial simulation	en
dc.title	Προσομοίωση της καταλυτικής πυρόλυσης με χρήση μηχανικής μάθησης και ενσωμάτωσης δεδομένων	el
dc.title	Data embedding and machine learning in modeling industrial fluid catalytic cracking reactors	en
heal.type	bachelorThesis
heal.classification	Process engineering	en
heal.language	el
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2023-03-10
heal.abstract	Fluid catalytic cracking is a both a critical and complex operation for refineries. Furthermore, the nature of the process (complex feedstock, complex sets of reactions and products) bears similarity with the complexity of emerging technologies in bio-renewables such as pyrolysis or hydrothermal liquefaction. FCC models are either linear and bilinear models adapted by regression or data-driven models. Instead, the paper explores machine-learning technologies to combine first-principle based models by embedding operational data from a real process. This artificial data, derived from equation-based models acts directly on neural network parameters, steering them in the right direction without the formation of new or updated input-output equations. The embedding provides for a systematic use of measurements form the plant while the first-principle based models offer a basis to explain the performance of the model. Early results illustrate very satisfactory accuracy, faster times and the opportunity to reduce the complexity of the process.	en
heal.advisorName	Kokossis, Antonis	en
heal.committeeMemberName	Karonis, Dimitrios	en
heal.committeeMemberName	Tsopelas, Fotios	en
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Χημικών Μηχανικών. Τομέας Ανάλυσης, Σχεδιασμού και Ανάπτυξης Διεργασιών και Συστημάτων (ΙΙ)	el
heal.academicPublisherID	ntua
heal.numberOfPages	127 σ.	el
heal.fullTextAvailability	false