Views
14
Downloands
2
Deve gezgin davranışı algoritması (CA) 2016 yılında Mohammed Khalid Ibrahim ve Ramzy Salim Ali tarafından önerilmiş, doğadan ilham alan bir meta-sezgiseldir. Bilimsel literatürde CA’ nın performansını ölçen birkaç çalışma bulunmaktadır. CA literatürde global optimizasyon ve mühendislik problemlerine uygulanmıştır. CA’ nın global optimizasyonda parçacık sürü optimizasyonu ve genetik algoritmadan daha iyi performans sergilediği gösterilmiştir. Buna karşın, bu algoritma gezgin satıcı probleminde olduğu gibi kombinatoryal optimizasyonda düşük kaliteli çözümler vermektedir. Bunun yanında, değiştirilmiş deve algoritması mühendislik alanında uygulanmıştır ve CS, PSO, CA’ dan daha iyi olduğu ortaya koyulmuştur. Bu sebeple, CA’ nın tur oluşturucu bir sezgiselle (En yakın komşu algoritması-NN) hibrid edilerek iyileştirilmesi ihtiyacı bulunmaktadır. Bu karşılaştırmalı uygulamada, 29-195 arasında değişen boyutlarda şehir içeren 13 küçük ve orta ölçekli veriseti kullanılmıştır. Sonuçlar, hibrid algoritmanın (HA), tabu arama (TS), GA, CA, ve karınca sistemine (AS) göre wi29, eil76, pr76, ve rat99 dışında bütün verisetlerinin %70 inde daha üstün olduğunu göstermektedir. Çalışmada, detaylı bir analiz verilerek en iyi, en kötü, ortalama çözümler, standard sapma, ve ortalama CPU zamanları sunulmaktadır. Metrikler, ayrıca hibrid meta-sezgiselin kabul edilebilir çözümleri bulmada 64% performans sergilediğini vurgulamaktadır. Sonuç olarak, hibrid algoritma küçük ve orta ölçekli verisetlerinde diğer test algoritmalarına kıyasla kesikli problemi daha kısa hesaplama zamanlarında çözmektedir.
The Deve traveler behavior algorithm (CA) is a meta-sessential, inspired by nature, proposed by Mohammed Khalid Ibrahim and Ramzy Salim Ali in 2016. There are several studies in scientific literature that measure the performance of CA. CA literature has been applied to global optimization and engineering issues. CA has been shown to perform better in global optimization than particle mass optimization and genetic algorithms. Nevertheless, this algorithm provides low-quality solutions in combinatory optimization, as is the problem of the traveler seller. In addition, the modified dust algorithm has been applied in the field of engineering and it has been shown to be better than CS, PSO, CA. Therefore, CA needs to be improved by hybridizing a tour-creating intuition (The Nearest Neighbor's Algorithm-NN). In this comparative application, 13 small and medium-sized data included cities in sizes ranging from 29 to 195 were used. The results show that the hybrid algorithm (HA), taboo search (TS), GA, CA, and mercury system (AS) are 70% superior to all data, except wi29, eil76, pr76, and rat99. In the study, a detailed analysis is provided with the best, worst, average solutions, standard deviation, and average CPU times. Metrics also emphasize that the hybrid meta-section shows 64% performance in finding acceptable solutions. As a result, hybrid algorithms solve the cut problem in smaller and medium-sized data compared to other test algorithms in shorter calculation times.
Camel Traveling Behavior Algorithm (CA) is a nature-inspired meta-heuristic proposed in 2016 by Mohammed Khalid Ibrahim and Ramzy Salim Ali. There exist few publications that measure the performance of the CA on scientific literature. CA was implemented to global optimization and some engineering problems in the literature. It was shown that CA demonstrates better performance than Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) in global optimization. However, it gives poor solutions at combinatorial optimization as well as in traveling salesman problems (TSP). Besides, a modified camel algorithm (MCA) was applied in the field of engineering and was proved that it is better than Cuckoo Search (CS), PSO, and CA. Therefore, it is a need for improvement in CA by hybridizing with a constructive heuristic (Nearest Neighbor Algorithm-NN). A set of thirteen small and medium-scale datasets that have cities scales ranging from 29 to 195 was used in the comparative study. The results show that the hybrid algorithm (HA) outperforms Tabu Search (TS), GA, CA, and Ant system (AS) for 70% of all datasets, excluding wi29, eil76, pr76, and rat99. Also, it was given that a detailed analysis presents the number of best, worst, average solutions, standard deviation, and average CPU time. The metrics also stress that the hybrid meta-heuristic demonstrates 64% performance in finding acceptable solutions. Finally, the hybrid algorithm solves the discrete problem in short computational times when compared to other test algorithms for small and medium-scale datasets.
Field : Mühendislik; Fen Bilimleri ve Matematik
Journal Type : Ulusal
Article Detail 
Similar Articles
Download PDF
Journal Information
Listen Paper
Cite
Print This Page
Share
Mendeley
