A Risk-based approach and computational Bayesian framework for off-line condition monitoring of marine diesel engine crankcase lubrication.

Αννέτης, Εμμανουήλ; Annetis, Emmanouil

dc.contributor.author	Αννέτης, Εμμανουήλ	el
dc.contributor.author	Annetis, Emmanouil	en
dc.date.accessioned	2020-10-14T07:12:19Z
dc.date.available	2020-10-14T07:12:19Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/51438
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.19136
dc.rights	Default License
dc.subject	Dynamic Bayesian networks	en
dc.subject	Risk analysis	en
dc.subject	Condition monitoring	en
dc.subject	Lubrication oil analysis	en
dc.subject	deterioration model	en
dc.subject	Prediction model	en
dc.subject	Ανάλυση ρίσκου	el
dc.subject	Ανάλυση λαδιού ναυτικού κινητήρα	el
dc.subject	Μοντέλα πρόβλεψης	el
dc.subject	Δυναμικά δίκτυα πίστεως	el
dc.subject	Παρακολούθηση κατάστασης ναυτικού κινητήρα	el
dc.subject	Μοντέλα γήρανσης	el
dc.title	A Risk-based approach and computational Bayesian framework for off-line condition monitoring of marine diesel engine crankcase lubrication.	en
heal.type	bachelorThesis
heal.classification	Marine engineering	en
heal.classification	Risk analysis	en
heal.classification	Bayesian networks	en
heal.classification	Probabilistic models	en
heal.classification	Ναυτική μηχανολογία	el
heal.classification	Μοντέλα πρόβλεψης	el
heal.classification	Ανάλυση ρίσκου	el
heal.language	en
heal.access	campus
heal.recordProvider	ntua	el
heal.publicationDate	2020-07-27
heal.abstract	Present thesis attempts an introduction to machine learning methodologies. This introduction is attempted along with the creation of a dynamic probabilistic predictive model which simulates deterioration trends. The general approach for such a model is a risk-based approach. Risk can be useful for decision making, as it considers the expected value of the degree of impact. The implementation of such a model is achieved through a dynamic probabilistic graphical model, namely a dynamic Bayesian belief network. Dynamic Bayesian networks are a powerful tool for risk modeling and for creating prediction models. The graphical representation of Bayesian networks makes them understandable and accessible even to non-familiar researchers or maritime executives. The above methodology is implemented on the lubrication of marine diesel engine, namely the main engine of large commercial vessels. Studies show that there are enough motivating reasons for this specific focus, as lubrication failures are the most significant cause of damage, both in number and average cost. Meanwhile, considering a risk-based approach for machinery systems constitutes a challenge, since they are not widely adopted. Thus, a dynamic Bayesian network is developed which simulates deterioration of engine components, oil degradation, and the off-line condition monitoring technique of oil analysis along with subsequent maintenance and repairs. Oil analysis is introduced to the system as the most common condition monitoring technique for marine engine lubrication. Additionally, an effective modeling of dependencies among lubricant properties, lubrication system parameters, failure mechanisms and failing components must be produced in order to achieve realistic simulations. Real data and respective analysis, along with expert elicitation and literature references are utilized for achieving model quantification. A variety of oil analysis interval schemes are defined, and a decision upon the most beneficial scheme follows, based on risk comparison. Matlab programming is used, along with the Bayes Net toolbox libraries, for producing the code of the model.	en
heal.advisorName	Βεντίκος, Νικόλαος	el
heal.advisorName	Ventikos, Nikolaos	en
heal.committeeMemberName	Βεντίκος, Νικόλαος	el
heal.committeeMemberName	Σαμουηλίδης, Εμμανουήλ	el
heal.committeeMemberName	Ηλιοπούλου, Ελευθερία	el
heal.committeeMemberName	Ventikos, Nikolaos	en
heal.committeeMemberName	Samouilidis, Emmanouil	en
heal.committeeMemberName	Iliopoulou, Eleftheria	en
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Ναυπηγών Μηχανολόγων Μηχανικών. Τομέας Μελέτης Πλοίου και Θαλάσσιων Μεταφορών	el
heal.academicPublisherID	ntua
heal.numberOfPages	159 σ.	el
heal.fullTextAvailability	false