Ģenētisko algoritmu papildināšana ceļojošā tirgoņa problēmas risināšanai
Date
2016
Authors
Ozoliņš, Edmunds
Journal Title
Journal ISSN
Volume Title
Publisher
Latvijas Universitāte
Abstract
Bakalaura darbā tiek risināta ceļojošā tirgoņa problēma, kurā ir jāatrod īsākais ceļš starp visām dotajām pilsētām, apmeklējot katru vienreiz un atgriežoties sākuma pilsētā. Darba mērķis ir papildināt ģenētiskos algoritmus, lai uzlabotu to spēju risināt ceļojošā tirgoņa problēmu. Darba uzdevumi ir izpētīt ģenētisko algoritmu papildināšanas iespējas, izveidot pamata risinājumu, veikt tā papildināšanu un novērtēt tā rezultativitāti izvirzītās problēmas risināšanā. Darba rezultātā tika izstrādāta struktūra, kas papildina ģenētiskos algoritmus ar Kohonen klasterizāciju, tuvākā kaimiņa algoritmu, 2-opt optimizāciju un stohastisko kalnā kāpšanu, kā arī tika iegūts izstrādāto papildinājumu novērtējums.
The bachelor’s thesis deals with the travelling salesman problem where the shortest route between all given cities must be found by visiting each once and returning to the starting city. The goal of the work is to supplement genetic algorithms to improve their ability to deal with the travelling salesman problem. The work objectives are to research genetic algorithm supplementing options, to create a base solution, to supplement it and to evaluate its effectiveness in dealing with the proposed problem. As a result, a structure was created that supplements genetic algorithms with Kohonen clusterization, nearest neighbor algorithm, 2-opt optimization and stochastic hill climbing. Also an evaluation of the created supplements was acquired.
The bachelor’s thesis deals with the travelling salesman problem where the shortest route between all given cities must be found by visiting each once and returning to the starting city. The goal of the work is to supplement genetic algorithms to improve their ability to deal with the travelling salesman problem. The work objectives are to research genetic algorithm supplementing options, to create a base solution, to supplement it and to evaluate its effectiveness in dealing with the proposed problem. As a result, a structure was created that supplements genetic algorithms with Kohonen clusterization, nearest neighbor algorithm, 2-opt optimization and stochastic hill climbing. Also an evaluation of the created supplements was acquired.
Description
Keywords
Datorzinātne , TSP , GA , klasterizācija , NN , 2-opt