Συνδυασμός μεθοδολογιών αιτίου-αιτιατού με μεθοδολογίες διαγραμμάτων συσχετισμών για την ανάπτυξη fuzzy cognitive maps απεικόνισης κινδύνων σε βιομηχανικό εξοπλισμό προς συντήρηση
Combination of cause & effect methodologies with relationship diagrams methodologies for the development of fuzzy cognitive maps for risk mapping in industrial equipment maintenance
View/
Keywords
Ασαφής γνωστικός χάρτης ; Δενδρικό μοντέλο ; Προγνωστική συντήρηση ; Εργασίες συντήρησης ; Χρονοδιάγραμμα ; Μηχανική μάθηση ; Δεξαμενή πετρελαίου ; Χρονική καθυστέρηση ; Προδιαγραφή συντήρησης ; Τεχνητή νοημοσύνηAbstract
This Master Thesis examines the combination of cause-and-effect methodologies with correlation diagram methodologies for the development of Fuzzy Cognitive Maps (FCMs) that depict risks that may arise following maintenance on industrial equipment, specifically in the field of petroleum tank maintenance.
The first chapter analyzes the various parameters that affect maintenance tasks, with emphasis on the impact of delays and their causes, as examined through Pareto analysis. Additionally, the benefits of predictive maintenance, the role of decision trees, and the contribution of machine learning are discussed.
Subsequently, the master thesis focuses on the mechanical structure of petroleum tanks, maintenance tasks, and scheduling, providing a comprehensive analysis of their structure. Moreover, it addresses the risks that may arise from incorrect maintenance procedures or failure to comply with maintenance standards.
Finally, a concise analysis of Fuzzy Cognitive Maps is presented, where FCMs are explored as a tool for understanding and managing risks associated with maintenance. The Hebbian learning method is also examined as a tool to improve the accuracy of these models.
The final conclusion of the diploma thesis is that the integration of machine learning leads to the creation of more efficient FCMs aimed at optimizing various processes.