Mühendislik problemlerinin çözümü için yeni bir hiper-sezgisel algoritma tasarımı
Küçük Resim Yok
Tarih
2024
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Düzce Üniversitesi
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Bu tezde, Hiper-Sezgisel Uygunluk-Mesafe Dengesi Başarı-Tarih Temelli Uyarlanabilir Diferansiyel Evrim (HH-FDB-SHADE) adı verilen yeni bir optimizasyon algoritması sunulmaktadır. Hiper-sezgisel algoritmalar iki ana yapıya sahiptir: Bir hiper seçim arayüzü ve bir düşük seviye sezgisel (LLH) havuzu. Önerilen algoritmada, LLH havuzu algoritmalarını değerlendirmek için yüksek seviyeli bir seçim arayüzü olarak FDB yöntemi tercih edilmiştir. Ayrıca, LLH havuzu olarak kullanılmak üzere beş mutasyon operatörü ve iki çaprazlama yönteminden toplam on farklı strateji türetilmiştir. FDB'nin keşif ve sömürü kabiliyetlerini dengede tutması, önerilen algoritmanın seçim arayüzü olmasının temel nedenidir. HH-FDB-SHADE algoritmasının başarısı, farklı boyutlu arama alanları için CEC-17 ve CEC-20 kıyaslama fonksiyonlarında test edilmiş ve HH-FDB-SHADE'den elde edilen çözümler on farklı LLH havuzu algoritmasıyla karşılaştırılmıştır. İstatistiksel analizlerin sonuçları, HH-FDB-SHADE'nin CEC-17 ve CEC-20 kıyaslama problemlerini çözmek için en iyi sırada yer alan algoritma olduğunu ve LLH havuz algoritmalarına kıyasla daha iyi sonuçlar verdiğini göstermektedir. Ayrıca, HH-FDB-SHADE algoritması, iyileştirilmiş algoritmanın performansını daha açık bir şekilde ortaya koymak ve mühendislik problemlerini çözmedeki başarısını kanıtlamak için iki farklı mühendislik problemine uygulanmıştır. İlk mühendislik probleminde Otomatik Gerilim Regülatörü (OGR) tasarımı için PID, PIDF, FOPID ve PIDD2 denetleyici parametrelerinin optimize edilmesi amaçlanmıştır. OGR sisteminden elde edilen sonuçlar, literatürdeki Uygunluk-Mesafe Dengesi Lévy Uçuş Dağılımı, Diferansiyel Evrim, Harris Hawks Optimizasyonu, Barnacles Çiftleşme Optimizasyonu ve Güve Alevi Optimizasyonu algoritmaları gibi beş diğer etkili meta-sezgisel arama algoritmasıyla karşılaştırılmıştır. Önerilen algoritma, optimal OGR tasarım problemlerini çözmede diğer beş meta-sezgisel algoritmadan daha etkili ve sağlam sonuçlar vermiştir. Bir diğer mühendislik probleminde ise Alternatif Akım/Çok Terminalli Doğru Akım Entegreli Optimal Güç Akışı (AC/MTDC OPF) için beş farklı amaç fonksiyonu optimize edilmiştir. Önerilen algoritma ile elde edilen sonuçlar, literatürde yer alan Uyarlamalı Yönlendirilmiş Diferansiyel Evrim, Deniz Yırtıcıları Algoritması, Atom Arama Optimizasyonu, Stokastik Fraktal Arama ve Uygunluk-Mesafe Dengesi tabanlı Stokastik Fraktal Arama algoritmaları ile karşılaştırılmıştır. Karşılaştırma sonucunda elde edilen bulgularda, önerilen algoritmanın AC/MTDC OPF problemi için diğer algoritmalara kıyasla etkili ve iyi sonuçlar verdiği gözlemlenmektedir.
This thesis proposed a new optimization algorithm based on the hyper-heuristic structure. The proposed optimization algorithm is called Hyper-Heuristic Fitness-Distance Balance Success-History Based Adaptive Differential Evolution (HH-FDB-SHADE). The hyper selection framework and a low-level heuristic (LLH) pool are the two main structure of the hyper-heuristic algorithms. The hyper selection framework selects the most suitable heuristic method among the low-level heuristic pool and this selection method was chosen to be the FDB in the proposed algorithm. Moreover, ten different heuristic methods which are obtained from two crossover methods and five mutation operators are chosen to create the LLH pool. The main reason for choosing FDB as the framework of the proposed algorithm is that it is a good method to balance exploration and exploitation capability. In order to test the success of HH-FDB-SHADE algorithm, CEC-17 and CEC-20 benchmark test suites in the literature are tested for different dimensional search spaces and the obtained results are compared with ten different LLH pool algorithms. The best ranked algorithm in solving CEC-17 and CEC-20 benchmark test suites is HH-FDB-SHADE according to the results of statistical analysis. In addition, the HH-FDB-SHADE algorithm is applied to two different engineering problems to reveal the performance of the improved algorithm more clearly and to prove its success in solving engineering problems. The first engineering problem is selected as the optimization of the control parameters of PID, PIDF, FOPID and PIDD2 in the Automatic Voltage Regulator (AVR) design problem. The results obtained from the AVR system are compared with five other effective meta-heuristic search algorithms in the literature such as Fitness-Distance Balance Lévy Flight Distribution, Differential Evolution, Harris Hawks Optimization, Barnacles Mating Optimization and Moth Flame Optimization algorithms. Also, the proposed algorithm is more effective and robust than the other five meta-heuristic algorithms in solving AVR design problems. As another engineering problem, the optimization of five different objective functions has been selected for the Alternating Current/Multi-Terminal Direct Current Integrated Optimal Power Flow (AC/MTDC OPF) problem. The results obtained with the proposed algorithm are compared with the Adaptive Guided Differential Evolution, Marine Predators Algorithm, Atom Search Optimization, Stochastic Fractal Search and Fitness-Distance Balance based Stochastic Fractal Search algorithms in the literature. According to the comparison result, it is observed that the proposed algorithm gives effective results compared to other algorithms for the AC/MTDC OPF problem.
This thesis proposed a new optimization algorithm based on the hyper-heuristic structure. The proposed optimization algorithm is called Hyper-Heuristic Fitness-Distance Balance Success-History Based Adaptive Differential Evolution (HH-FDB-SHADE). The hyper selection framework and a low-level heuristic (LLH) pool are the two main structure of the hyper-heuristic algorithms. The hyper selection framework selects the most suitable heuristic method among the low-level heuristic pool and this selection method was chosen to be the FDB in the proposed algorithm. Moreover, ten different heuristic methods which are obtained from two crossover methods and five mutation operators are chosen to create the LLH pool. The main reason for choosing FDB as the framework of the proposed algorithm is that it is a good method to balance exploration and exploitation capability. In order to test the success of HH-FDB-SHADE algorithm, CEC-17 and CEC-20 benchmark test suites in the literature are tested for different dimensional search spaces and the obtained results are compared with ten different LLH pool algorithms. The best ranked algorithm in solving CEC-17 and CEC-20 benchmark test suites is HH-FDB-SHADE according to the results of statistical analysis. In addition, the HH-FDB-SHADE algorithm is applied to two different engineering problems to reveal the performance of the improved algorithm more clearly and to prove its success in solving engineering problems. The first engineering problem is selected as the optimization of the control parameters of PID, PIDF, FOPID and PIDD2 in the Automatic Voltage Regulator (AVR) design problem. The results obtained from the AVR system are compared with five other effective meta-heuristic search algorithms in the literature such as Fitness-Distance Balance Lévy Flight Distribution, Differential Evolution, Harris Hawks Optimization, Barnacles Mating Optimization and Moth Flame Optimization algorithms. Also, the proposed algorithm is more effective and robust than the other five meta-heuristic algorithms in solving AVR design problems. As another engineering problem, the optimization of five different objective functions has been selected for the Alternating Current/Multi-Terminal Direct Current Integrated Optimal Power Flow (AC/MTDC OPF) problem. The results obtained with the proposed algorithm are compared with the Adaptive Guided Differential Evolution, Marine Predators Algorithm, Atom Search Optimization, Stochastic Fractal Search and Fitness-Distance Balance based Stochastic Fractal Search algorithms in the literature. According to the comparison result, it is observed that the proposed algorithm gives effective results compared to other algorithms for the AC/MTDC OPF problem.
Açıklama
Anahtar Kelimeler
Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
