Ekonomik ve minimum emisyon dağıtımı
Yükleniyor...
Dosyalar
Tarih
item.page.authors
Süreli Yayın başlığı
Süreli Yayın ISSN
Cilt Başlığı
Yayınevi
Fen Bilimleri Enstitüsü
Institute of Science and Technology
Institute of Science and Technology
Özet
Son yıllarda, artan çevre bilincine paralel olarak enerji sistemlerinde, üretim sonucunda ortaya çıkan emisyonlar da önem kazanmış ve üretim planı yapılırken maliyetle birlikte emisyon da hesaplara katılmıştır. Bu tez çalışmasının amacı ekonomik ve minimum emisyon dağıtımını, farklı optimizasyon teknikleriyle gerçekleştirmektir. Bahsedilen bu optimizasyon teknikleri, Hopfield Neural Network (HNN), Lagrange Çarpanları (LM) ve Quick Method (QM)' dur. Bu optimizasyon tekniklerinin yanında, Hopfield Neural Network yöntemindeki diferansiyel denklemin çözümü için de Açık Euler ve Düzeltilmiş Euler yöntemi ile Runge Kutta- 4 yönteminden yararlanılmıştır. Ayrıca, Hopfield Neural Network yönteminde, optimuma daha hızlı yaklaşmak için momentum teriminden de yararlanılmıştır. Ekonomik ve minimum emisyon dağıtımı probleminin modeli ise, bu konuda daha önce yapılmış çalışmaların ışığında, her generatörün aktif güç değerinin ikinci dereceden fonksiyonu şeklinde kabul edilen yakıt maliyeti ve NOx emisyonu ifadelerinin, ağırlık katsayılarından yararlanılarak bir amaç fonksiyonu altında bMeştirilmesiyle elde edilmiştir. Ekonomiklik ve minimum emisyon şeklinde iki amaca sahip problem için, 6 generatörlü bir sistemde, bu 3 optimizasyon tekniğiyle, sadece ekonomik dağıtım, sadece emisyon dağıtımı ve bu iki amacın birlikte kullanıldığı ekonomik ve minimum emisyon dağıtımı, 500, 600 ve 700 MW talep güçleri için, kayıpsız ve iletim kayıplarının hesaba katıldığı iki ayrı durum için yapılmıştır. Simülasyon bölümünde detaylı olarak açıklandığı gibi, ilk olarak Lagrange Çarpardan ile diğer yöntemler karşılaştrrılmıştır. Daha sonra da, diferansiyel denkleminin çözümünde hangi tekniğin daha iyi olduğunu anlamak için Düzeltilmiş Euler kullanılan Hopfield Neural Network ile diğer teknikleri kullanan Hopfield Neural Network' 1er karşılaştınlmıştrr. Genel itibariyle, hata değerleri literatürdeki çalışmalara çok yakındır ve başarım oranlan ise oldukça yüksektir. Lagrange Çarpardan ve Quick Method gibi klasik yöntemlerin yanında, Hopfield Neural Network de iyi sonuçlar vermiştir. Aynca, Hopfield diferansiyel denklemi için kullanılan diferansiyel denklem çözümleri içinde, en iyisi Runge Kutta-4 ile momentumun birlikte kullanıldığı durumdur. Sonuç olarak, tüm tekniklerle iyi sonuçlann elde edildiği söylenebilir.
In the last years, while environment conscious has been increasing, emissions which come into being after production have won importance and while production plan has been done, emission has been added to calculation together with cost in the energy systems. The objective of this thesis is to do economic and minimum emission dispatch with different optimization techniques. These optimization techniques are Hopfield Neural Network, Lagrange Multiplier, and Quick Method. Apart from these, Improved Euler Method and Runge Kutta-4 Method are used to solve differential equation in Hopfield Neural Network Method. Also, in Hopfield Neural Network Method, momentum term is used to converge to optimum point quickly. The model of economic and minimum emission dispatching problem is, at the light of previous studies, built up putting together fuel cost and NOx emission functions, which are accepted as the second order function of active power value of every generator, using weight coefficients under the objective function. In this study, only economic dispatch, only emission dispatch, and economic and minimum emission dispatch was employed for the problem which has economic and minimum emission objective, in a system with six generators, with these three optimization techniques, for 500, 600, and 700 MW demand powers, for the two conditions without transmission loss and with transmission loss. Firstly, Lagrange Multiplier method is compared with other methods as explained in simulation section in detail. Then, Hopfield Neural Network with Improved Euler Method is compared with Hopfield Neural Network Methods with other differential equation solution techniques to understand which technique is the best in differential equation solution techniques. The fault values of the results are near the same as in the literature. The performance of Hopfield Neural Network is very good, so this method can be comparable to the classical methods. Also, the condition used together Runge Kutta- 4 Method and momentum term is the best in differential equation solution techniques used for Hopfield differential equation. In general these three optimization techniques showed a promising result to implement economic and minimum emission dispatch.
In the last years, while environment conscious has been increasing, emissions which come into being after production have won importance and while production plan has been done, emission has been added to calculation together with cost in the energy systems. The objective of this thesis is to do economic and minimum emission dispatch with different optimization techniques. These optimization techniques are Hopfield Neural Network, Lagrange Multiplier, and Quick Method. Apart from these, Improved Euler Method and Runge Kutta-4 Method are used to solve differential equation in Hopfield Neural Network Method. Also, in Hopfield Neural Network Method, momentum term is used to converge to optimum point quickly. The model of economic and minimum emission dispatching problem is, at the light of previous studies, built up putting together fuel cost and NOx emission functions, which are accepted as the second order function of active power value of every generator, using weight coefficients under the objective function. In this study, only economic dispatch, only emission dispatch, and economic and minimum emission dispatch was employed for the problem which has economic and minimum emission objective, in a system with six generators, with these three optimization techniques, for 500, 600, and 700 MW demand powers, for the two conditions without transmission loss and with transmission loss. Firstly, Lagrange Multiplier method is compared with other methods as explained in simulation section in detail. Then, Hopfield Neural Network with Improved Euler Method is compared with Hopfield Neural Network Methods with other differential equation solution techniques to understand which technique is the best in differential equation solution techniques. The fault values of the results are near the same as in the literature. The performance of Hopfield Neural Network is very good, so this method can be comparable to the classical methods. Also, the condition used together Runge Kutta- 4 Method and momentum term is the best in differential equation solution techniques used for Hopfield differential equation. In general these three optimization techniques showed a promising result to implement economic and minimum emission dispatch.
Açıklama
Tez (Yüksek Lisans) -- İstanbul Teknik Üniversitesi, Fen Bilimleri Enstitüsü, 2003
Thesis (M.Sc.) -- İstanbul Technical University, Institute of Science and Technology, 2003
Thesis (M.Sc.) -- İstanbul Technical University, Institute of Science and Technology, 2003
Konusu
Elektrik gücü sistemleri, Elektrik gücü iletimi, Elektrik gücü dağıtımı, Electric power systems, Electric power transmission, Electric power distribution
