Solving Generalized Traveling Salesman Problem by Using Discrete Differantial Evaluation Algorithm Hibrydized Wıth Local Search Heurıstic
Loading...

Date
2010
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Open Access Color
OpenAIRE Downloads
OpenAIRE Views
Abstract
Bu tezde genellestirilmis gezgin satıcı probleminin çözümü için yerel tarama ile birlestirilmis kesikli farksal evrim algoritması sunulmustur. Genellestirilmis gezgin satıcı probleminde bir satıcının is yaptıgı sehirler kümelere ayrılır ve satıcının her kümeden yalnız bir sehre ugrayarak en kısa yoldan turu tamamlaması beklenir.Bu algoritmayı test etmek için, GTSPLIB kütüphanesinde bulunan, sehir ve küme sayıları 48 (10) ile 1084 (217) arasında degisen 54 test problemi kullanılmıstır.Sonuçların deneysel analizlerinin yapılması ile, algoritma en iyi sonuçları veren Bontoux, Artigues ve Feillet'in (2009) Memetik Algoritması, Tasgetiren, Suganthan ve Pan'ın (2009) eDDE algoritması, Synder ve Daskin'in (2006) RKGA veSilberholz ve Golden'ın (1997), mrOXGA ile kıyaslanmıstır. Sonuç olarak, eniyi degerleri bilinen 41 test probleminin sonuçları kıyaslandıgında, KFE Algoritması mrOXGA, MA ve eDDE algoritmasına esdeger oldugu ancak RKGA'dan daha iyi sonuçlar ürettigi görülmüstür.
This thesis presents a discrete differential evaluation algorithm hybridized with a local search heuristic (KFE), for the generalized traveling salesman problem. In the GTSP, the set of cities is divided into clusters so that the aim is to find minimum tour length when a salesman has to visit one city from every cluster.In order to test this algorithm, 54 benchmark instances ranging from 48 (10) to 1084(217) nodes/ clusters from the GTSPLIB are used. Through the experimental analysis of the results, the performance of the algorithm is compared against the best performing algorithms such as Memetic Algorithm of Bontoux, Artigues and Feillet,eDDE algorithm of Tasgetiren, Suganthan and Pan, RKGA of Synder and Daskin,and mrOXGA of Silberholz and Golden.
This thesis presents a discrete differential evaluation algorithm hybridized with a local search heuristic (KFE), for the generalized traveling salesman problem. In the GTSP, the set of cities is divided into clusters so that the aim is to find minimum tour length when a salesman has to visit one city from every cluster.In order to test this algorithm, 54 benchmark instances ranging from 48 (10) to 1084(217) nodes/ clusters from the GTSPLIB are used. Through the experimental analysis of the results, the performance of the algorithm is compared against the best performing algorithms such as Memetic Algorithm of Bontoux, Artigues and Feillet,eDDE algorithm of Tasgetiren, Suganthan and Pan, RKGA of Synder and Daskin,and mrOXGA of Silberholz and Golden.
Description
Keywords
Evrimsel Algoritmalar, Gezgin Satıcı Problemi, Travelling Salesman Problem, İşletme, Business Administration, Evolutionary Algorithms
Turkish CoHE Thesis Center URL
Fields of Science
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
End Page
70
