Karesel atama problemi çok farklı alanlarda farklı tipleriyle karşılaşılabilen bir problem türüdür. Birbiri arasında akış olan faaliyet noktalarının birbiri arasında belirli mesafe olan lokasyonlara atanmasına dayanan ve akış ile mesafenin birlikte ele alınması gereken problemin, klasik çözüm yöntemleri ile çözümü oldukça zor olmakta ve çoğunlukla sezgisel algoritmalar ile uygun çözümler elde edilebilmektedir. Çok sık kullanılan sezgisel algoritmalarından biri ise Genetik Algoritma’dır. Genetik Algoritma rassallık içeren prosedürlere sahip, canlıların evrimini taklit eden oldukça başarılı bir yöntemdir. Algoritmanın en önemli prosedürlerinden bir tanesi ise çaprazlama işlemidir. Karesel atama problemi için literatürde sıklıkla pozisyon temelli çaprazlama tercih edilmektedir. Pozisyon temelli çaprazlamada kaç adet noktanın sabit kalacağı ve bu değerin nasıl belirleneceği önemli bir durumdur. Çalışmada, sabit tutulan nokta sayısını önceden belirli değerler ile ele alan klasik Genetik Algoritma yaklaşımları ile önerilen yöntem olan bulanık adaptif yaklaşım kıyaslanmıştır. Yapay zekanın bir türü olan Bulanık Mantık teorisinden faydalanan bulanık adaptif yaklaşım sayesinde algoritmanın çözüm esnasında elde ettiği bilgiler kullanılarak parametreler kontrol edilmekte ve algoritmanın arama yönü daha akıllı bir şekilde değişebilmektedir. Önerilen yöntemin etkinliğini değerlendirebilmek için literatürde yer alan karesel atama problem örneklerinden yararlanılmıştır. Sonuçların değerlendirilmesi ile Genetik Algoritma’da bulanık adaptif yaklaşımın etkinliği ortaya çıkmıştır.
A carrier assignment problem is a type of problem that can be faced with different types in very different fields. Based on the assignment of activity points that flow between each other to locations that have a certain distance between each other and the problem that flow and distance must be dealt with together is quite difficult to solve with the classic solutions methods and mostly with intuitive algorithms the appropriate solutions can be achieved. One of the most commonly used algorithms is the genetic algorithm. The genetic algorithm is a very successful method of imitating the evolution of animals with procedures that include raciality. One of the most important procedures of the algorithm is the cross-process. In literature for the problem of carrier assignment, position-based crossover is often preferred. It is important how many points will remain stable in the position-based crossover and how this value will be determined. The study compared the foolish adaptive approach, which is the recommended method with the classic genetic algorithm approaches that deal with the number of fixed points with predefined values. The foolish adaptive approach, which is a kind of artificial intelligence, is the theory of the Bully Logic, which uses the information obtained by the algorithm during the solution to control the parameters and the search direction of the algorithm can be changed more intelligently. For evaluating the effectiveness of the proposed method, the examples of the quarterly assignment problem found in the literature have been used. The evaluation of the results revealed the effectiveness of a foolish adaptive approach in the Genetic algorithm.
The quadratic assignment problem is a type of problem that can be encountered in many different fields. The problem, which is based on the assignment of the points of activity which have flow between each other to locations with a certain distance between each other and where flow and distance has to be handled together, is very difficult to solve by classical solution methods and mostly suitable solutions can be obtained with heuristic algorithms. One of the most frequently used heuristic algorithms is Genetic Algorithm. Genetic Algorithm is a very successful method that simulates the evolution of living things with procedures involving randomness. One of the most important procedures of the algorithm is crossover. For the quadratic assignment problem, position-based crossover is often preferred in the literature. How many points will remain constant and how to determine this value is important for position based crossover. In this study, the classical Genetic Algorithm approaches, which treat the number of fixed points with predetermined values, are compared with the proposed method, fuzzy adaptive approach. By using the fuzzy adaptive approach which uses Fuzzy Logic theory which is a kind of artificial intelligence, the parameters are controlled by using the information obtained by the algorithm during the solution and the search direction of the algorithm can change more intelligently. In order to evaluate the effectiveness of the proposed method, the examples of quadratic assignment problems in the literature were utilized. With the evaluation of the results, the effectiveness of the fuzzy adaptive approach in Genetic Algorithm was revealed.
Alan : Sosyal, Beşeri ve İdari Bilimler
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|