University of the Aegean
Νοήμονες υπολογιστικές μέθοδοι εμπνευσμένες από τον φυσικό κόσμο για την βελτιστοποίηση συστημάτων: βελτιστοποίηση εμπνευσμένη από τον ηχοεντοπισμό
Abstract
dc:descriptionThe current era brings Artificial Intelligence to the forefront, as various industries incorporate impressive applications. Many references are made to the 4th Industrial Revolution, which now has Artificial Intelligence (AI) as its main pillar. As early as the 1950s, intelligent decision-making systems were being developed, which usually played a supporting role for a decision-maker. Problems in which an optimal strategy had to be found, the so-called optimization problems, arise in a variety of areas such as Financial Engineering, Management Engineering and Operational Research. Many of the problems that an engineer has to solve are optimization problems. While the problems that the industry faces become more and more difficult, the usefulness of Intelligent Computing and its integration into the relevant industry sectors is increasing. The range of Computational Intelligence oriented methods is wide, ranging from Machine Learning to Nature-inspired Intelligent (NII) Methods. Although NII methods initially appeared back in the mid-80s, they constituted a separate category of Computational Intelligent (CI) only in the early 2000's. Due to their ability to solve difficult multi-dimensional optimization problems, Computational Intelligence oriented methods have attracted the interest of researchers over the past decade. In this thesis, all Computational Intelligence oriented methods inspired by nature have been collected and studied in order to identify the features that can improve their performance. Moreover, this study focuses also on the application areas that these methods are used. As a result from this comprehensive literature review, some conclusions have been drawn. These conclusions constitute a framework that can be carried out for the development or improvement of Nature-inspired Intelligent Method. In addition, based on this framework and the aforementioned features, a new Nature-Inspired Intelligent approach inspired by the laws of underwater acoustics, has been developed. This approach has been proved to be effective in various application domains (Portfolio Optimization, Resource Leveling in Project Management, Economic Load Dispatch, Production Lines Optimization). What is more, compared with other state-of-the-art methods of literature and other Nature-Inspired Algorithms, the proposed approach obtains a superior performance. Throughout the conducted research, the main mechanisms of the proposed algorithm have been improved, based on the experimentation in the aforementioned application areas. Furthermore, some novel mechanisms (in respect of auto-tuning and balancing exploration and exploitation) have been implemented in the proposed approach. These mechanisms are part of the contribution of this thesis, since they can be implemented also in several Computational Intelligence oriented methods, in order to overcome problems that these methods usually face off. Furthermore, in some cases, hybrid intelligent schemes have been developed containing the proposed approach, in order to obtain improved performance.
Degree
thesis:*- Grantor dc:publisher
- University of the Aegean
- Year dc:date
- 2020
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Tzanetos, Alexandros
- Τζανέτος, Αλέξανδρος
Subjects
dc:subject × 12- Νοήμονες αλγόριθμοι εμπνευσμένοι από τη φύση
- Νοημοσύνη εμπνευσμένη από τη φύση
- Βελτιστοποίηση
- Nature inspired intelligent algorithms
- Nature inspired intelligence
- Optimization
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Άλλες Επιστήμες Μηχανικού και Τεχνολογίες
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
- Other Engineering and Technologies
Rights
dc:rights- Statement dc:rights
-
- BY_NC_SA
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/47419