Aristotle University Of Thessaloniki (AUTH)
Αξιολόγηση της επίδοσης πολυπρακτορικών συστημάτων σε εφαρμογές διαχείρισης εφοδιαστικής αλυσίδας
Abstract
dc:descriptionSoftware agents and multi-agent systems is a relatively young research area, which has already several applications in various fields. One such application is supply chain management, where multi-agent systems constitute an ideal solution, being able to easily model the complexity of the system and at the same time incorporating a considerable degree of intelligence and adaptivity. Performance in such systems is usually measured in terms of increased profit, which may be both the required result and an easily measurable quantity. This approach is relatively short-sighted and can be related to the structure and the essence of the multi-agent system only through experimentation. The evaluation of the multi-agent system as a software system is usually not addressed and the system's performance is rarely evaluated in depth. In this thesis, we present a generalized and reusable methodology to design and evaluate supply chain management multi-agent systems. We cover all the development stages of a system, from its design and knowledge incorporation, to its evaluation. First, we present a successful methodology for the design of supply chain-management multi-agent systems enhanced with data mining techniques. In a dynamic supply chain management environment, we took into account standard multi-agent system development practices and identified the parameters that best match our model. Using these parameters, we extensively analyzed the way in which data mining can be applied in order to enhance the intelligence of a software agent. Then, we introduce a novel method which enables us to make the best out of traditional data mining algorithms for time series data, where each sample depends on the previous ones, by overcoming dimensionality problems. We applied our method on two different case studies. Our method was capable of producing successful models in both cases, by using different algorithms. Finally, we offer an extensive set of metrics, which enables us to quantify and correctly evaluate different facets of the multi-agent system and evaluate its performance. These metrics are both agent-oriented and supply chain-oriented, in order to provide an accurate, global view of the system's performance. This methodology is demonstrated through its application on an existing open source multi-agent platform which simulates the function of the supply chain.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2011
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Nikolaidou, Vivia
- Νικολαΐδου, Παρασκευή
Subjects
dc:subject × 20- Πράκτορες λογισμικού
- Πολυπρακτορικά συστήματα
- Διαχείριση εφοδιαστικής αλυσίδας
- Ευφυή συστήματα
- Εξόρυξη γνώσης
- Αξιολόγηση συστημάτων
- Εξόρυξη χρονοσειρών
- Συστήματα μέτρησης
- Software agents
- Multi-agent systems
- Supply chain management
- Intelligent systems
- Knowledge mining from data
- Information systems evaluation
- Time series mining
- Measurement systems
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Engineering and Technology
- Electrical Engineering, Electronic Engineering, Information Engineering
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/28910