Uygunluk mesafe dengesi tabanlı sezgisel optimizasyon algoritmalarının güç sistemi problemlerine uygulanması
Loading...
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Düzce Üniversitesi
Access Rights
info:eu-repo/semantics/openAccess
Abstract
Bu tez çalışmasında, modern güç sistemlerinin planlanması ve işletiminde kritik öneme sahip kısıtlı güç sistemi problemlerinin çözümü için hibrit optimizasyon algoritmalarının geliştirilmesi amaçlanmıştır. Bu kapsamda, optimizasyon alanında iki önemli konu üzerinde araştırma yapılmış ve önemli kazanımlar elde edilmiştir. Bu konulardan ilki, uygunluk-mesafe dengesi (Fitness-Distance Balance, FDB) ve dinamik uygunluk-mesafe dengesi (Dynamic Fitness Distance-Balance, dFDB) seçim yöntemlerini kullanarak tek amaçlı meta-sezgisel algoritmaların arama performansının geliştirilmesidir. Erken yakınsama ve zayıf çeşitlilik problemlerinden muzdarip olan uyarlanabilir bilgi edinme-paylaşma (Adaptive Gaining-Sharing Knowledge, AGSK) ve Levy uçuş dağılımı (Levy Flight Distribution, LFD) algoritmalarının handikaplarını ortadan kaldırmak ve böylece arama performansını iyileştirmek için FDB seçim yöntemi kullanılmıştır. Baz algoritmaların seçim stratejileri FDB yöntemi kullanılarak yeniden tasarlanmış ve global optimizasyon problemlerini etkin bir şekilde çözme yeteneğine sahip hibrit FDBAGSK ve FDBLFD algoritmaları geliştirilmiştir. Bu çalışmanın orijinal katkılarından biri olarak geliştirilen dFDB seçim yöntemi manta ışını yiyecek arama optimizasyon (Manta Ray Foraging Optimization, MRFO) algoritmasına uygulanmış ve dFDB-MRFO olarak isimlendirilen yeni bir hibrit algoritma önerilmiştir. Araştırmanın yürütüldüğü ikinci konu, kısıtlı güç sistemi problemlerinin optimizasyonudur. Bu kapsamda ilk olarak, stokastik rüzgâr, güneş, hidro ve gelgit enerji kaynakları ile Çok-Terminalli Yüksek Gerilim Doğru Akım (ÇTYGDA) iletim hattı bağlantılarını içeren IEEE 30-baralı güç sisteminde Çok-Amaçlı Optimal Güç Akışı (ÇAOGA) probleminin optimizasyon modeli sunulmuştur. Önerilen problemin optimizasyonunda çok amaçlı çekirge optimizasyon algoritması (Multi-objective Grasshopper Optimization Algorithm, MOGOA) ve literatürdeki güçlü optimizasyon yöntemleri kullanılmıştır. Simülasyon sonuçları, MOGOA yönteminin, elde edilen Pareto optimal çözümlerin doğruluğu ve bunların dağılımı açısından oldukça rekabetçi sonuçlar elde ettiğini göstermiştir. Ardından, önerilen FDBAGSK algoritması dağıtılmış üretim ve iki terminalli YGDA iletim bağlantılarını içeren IEEE 30 ve IEEE 57-baralı güç sistemlerinde Alternatif Akım/Doğru Akım Optimal Reaktif Güç Akışı (AA/DA ORGA) probleminin çözümüne uygulanmıştır. Simülasyon sonuçlarından elde edilen bulgular, hibrit FDBAGSK algoritmasının büyük ölçekli güç sistemlerinde ORGA probleminin farklı konfigürasyonlarını verimli bir şekilde çözebildiğini göstermiştir. Optimal Otomatik Gerilim Regülatörü (OGR) tasarımında kullanılan PID, PIDF, FOPID ve PIDD2 kontrolcü parametrelerinin optimizasyonuna hibrit FDBLFD algoritması uygulanmıştır. Optimum OGR tasarımı için en iyi kontrolcü performansı, FDBLFD tabanlı PIDD2 ile elde edilmiştir. Son olarak, Yönlü Aşırı Akım Röleleri (YAAR) koordinasyon problemi hibrit dFDB-MRFO algoritması kullanılarak optimize edilmiştir. Önerilen optimizasyon yönteminin etkinliğini doğrulamak için, farklı karmaşıklığa sahip beş test sistemi üzerinde kapsamlı bir simülasyon gerçekleştirilmiştir. Simülasyon sonuçları, önerilen hibrit algoritmanın, YAAR koordinasyon problemini çözmek için verimli ve güvenilir bir yöntem olduğunu doğrulamıştır. Tüm sonuçlar bir arada düşünüldüğünde tez çalışmasında önerilen uygunluk mesafe dengesi tabanlı hibrit optimizasyon algoritmalarının güç sistemi problemlerine karşılaştırılan diğer algoritmalara kıyasla daha düşük hata ve yüksek doğruluk ile optimum çözümler üretebildiği gözlemlenmiştir.
In this thesis study, it is aimed to develop hybrid optimization algorithms for the solution of constrained power system problems that are critical in the planning and operation of modern power systems. In this context, research was carried out on two important issues in the field of optimization and significant achievements were obtained. The first of these issues was to improve the search performance of single-objective meta-heuristic algorithms by using Fitness-Distance Balance (FDB) and Dynamic Fitness-Distance Balance (dFDB) selection methods. FDB selection method was used to eliminate the handicaps of Adaptive Gaining-Sharing Knowledge (AGSK) and Levy Flight Distribution (LFD) algorithms, which suffer from premature convergence and poor diversity, and thus improve the search performance. The selection strategies of the original algorithms were redesigned using the FDB method, and thus hybrid FDBAGSK and FDBLFD algorithms were developed, which have the ability to effectively solve global optimization problems. The dFDB selection method, which was developed as one of the original contributions of this study, was applied to the Manta Ray Foraging Optimization (MRFO) algorithm and a new hybrid algorithm called dFDB-MRFO was proposed. The second subject on which research was conducted was the optimization of constrained power system problems. In this regard, firstly, the optimization model of the Multi-Objective Optimal Power Flow (MO-OPF) problem in IEEE 30-bus power system incorporating stochastic wind, solar, hydro, tidal energy sources and Multi-Terminal High Voltage Direct Current (MTHVDC) transmission links were presented. In the optimization of the proposed problem, the Multi-objective Grasshopper Optimization Algorithm (MOGOA) and the powerful optimization methods in the literature are used. The simulation results showed that the MOGOA method obtained very competitive results in terms of the accuracy of the obtained Pareto optimal solutions and their distribution. Then, the FDBAGSK algorithm is applied to the solution of the Alternating Current/Direct Current Optimal Reactive Power Flow (AC/DC ORPF) problem in IEEE 30 and IEEE 57-bus power systems incorporating distributed generation and two-terminal HVDC transmission links. Findings from the simulation results showed that the proposed FDBAGSK algorithm can efficiently solve different configurations of the ORPF problem in large-scale power systems. After that, the hybrid FDBLFD algorithm was applied to the optimization of the PID, PIDF, FOPID and PIDD2 controller parameters used in the optimal design of the Automatic Voltage Regulator (AVR). The best controller performance for optimal AVR design was achieved by FDBLFD based PIDD2. Finally, the Directional Overcurrent Relays (DOCRs) coordination problem was optimized using the hybrid dFDB-MRFO algorithm. To verify the effectiveness of the proposed optimization method, a comprehensive simulation was performed on five test systems with different complexities. The simulation results confirmed that the proposed hybrid algorithm is an efficient and reliable method to solve the DOCRs coordination problem. When all the results are considered together, it has been observed that the fitness distance balance-based hybrid optimization algorithms proposed in the thesis study can produce optimum solutions with lower error and high accuracy compared to other algorithms for power system problems.
In this thesis study, it is aimed to develop hybrid optimization algorithms for the solution of constrained power system problems that are critical in the planning and operation of modern power systems. In this context, research was carried out on two important issues in the field of optimization and significant achievements were obtained. The first of these issues was to improve the search performance of single-objective meta-heuristic algorithms by using Fitness-Distance Balance (FDB) and Dynamic Fitness-Distance Balance (dFDB) selection methods. FDB selection method was used to eliminate the handicaps of Adaptive Gaining-Sharing Knowledge (AGSK) and Levy Flight Distribution (LFD) algorithms, which suffer from premature convergence and poor diversity, and thus improve the search performance. The selection strategies of the original algorithms were redesigned using the FDB method, and thus hybrid FDBAGSK and FDBLFD algorithms were developed, which have the ability to effectively solve global optimization problems. The dFDB selection method, which was developed as one of the original contributions of this study, was applied to the Manta Ray Foraging Optimization (MRFO) algorithm and a new hybrid algorithm called dFDB-MRFO was proposed. The second subject on which research was conducted was the optimization of constrained power system problems. In this regard, firstly, the optimization model of the Multi-Objective Optimal Power Flow (MO-OPF) problem in IEEE 30-bus power system incorporating stochastic wind, solar, hydro, tidal energy sources and Multi-Terminal High Voltage Direct Current (MTHVDC) transmission links were presented. In the optimization of the proposed problem, the Multi-objective Grasshopper Optimization Algorithm (MOGOA) and the powerful optimization methods in the literature are used. The simulation results showed that the MOGOA method obtained very competitive results in terms of the accuracy of the obtained Pareto optimal solutions and their distribution. Then, the FDBAGSK algorithm is applied to the solution of the Alternating Current/Direct Current Optimal Reactive Power Flow (AC/DC ORPF) problem in IEEE 30 and IEEE 57-bus power systems incorporating distributed generation and two-terminal HVDC transmission links. Findings from the simulation results showed that the proposed FDBAGSK algorithm can efficiently solve different configurations of the ORPF problem in large-scale power systems. After that, the hybrid FDBLFD algorithm was applied to the optimization of the PID, PIDF, FOPID and PIDD2 controller parameters used in the optimal design of the Automatic Voltage Regulator (AVR). The best controller performance for optimal AVR design was achieved by FDBLFD based PIDD2. Finally, the Directional Overcurrent Relays (DOCRs) coordination problem was optimized using the hybrid dFDB-MRFO algorithm. To verify the effectiveness of the proposed optimization method, a comprehensive simulation was performed on five test systems with different complexities. The simulation results confirmed that the proposed hybrid algorithm is an efficient and reliable method to solve the DOCRs coordination problem. When all the results are considered together, it has been observed that the fitness distance balance-based hybrid optimization algorithms proposed in the thesis study can produce optimum solutions with lower error and high accuracy compared to other algorithms for power system problems.
Description
Keywords
Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
