![Institutional Repository [SANDBOX]](images/dias.logo.png "Institutional Repository [SANDBOX]"){kind=logo}
Institutional Repository [SANDBOX]
Technical University of Crete
Technical University of Crete
EN
|
EL
| URI: | http://purl.tuc.gr/dl/dias/CF06D423-5E47-45E3-888A-F221A46CBCD5 | ||
| Year | 2015 | ||
| Type of Item | Master Thesis | ||
| License |
|
||
| Bibliographic Citation | Νικόλαος Μακρυμανωλάκης, "Ανάπτυξη εξελικτικού αλγόριθμου και βελτιστοποίηση παραμέτρων λειτουργίας του για προβλήματα χρονοπρογραμματισμού εργασιών", Μεταπτυχιακή Διατριβή, Σχολή Μηχανικών Παραγωγής και Διοίκησης, Πολυτεχνείο Κρήτης, Χανιά, Ελλάς, 2015 https://doi.org/10.26233/heallink.tuc.26700 | ||
| Appears in Collections |
In this thesis we develop a new evolutionary algorithm, suitable for solving combinatorial optimization scheduling problems, such as jobs scheduling problems, the travelling salesperson problem, etc. We focus our study on flow-shop scheduling problem, where a number of n jobs has to be processed by a number of m machines, at the same sequence for every job, and we seek for smallest completion time. The algorithm combines various techniques used in local search. As various elements of the algorithm can be tuned, we follow a systematic data mining procedure and we utilize data from a number of executions in real problems, in order to seek models for the suitable parameterization for every m x n problem size. Finally we present our results using the model suggested parameters and we verify the quality of our results. The results show that the procedure we follow creates new promises on the improvement in efficiency for evolutionary algorithms in combinatorial optimization problems, and also in other applications where the fine-tuning of algorithms is important element in order to increase their efficiency.
Copyright © DIAS 2013
