Comparison of successes of different hybrid method formation strategies in short-term solar radiation prediction

Nazmiye Bozok; Emre Akarslan; Said Mahmut Çınar

Makale detayları

Comparison of successes of different hybrid method formation strategies in short-term solar radiation prediction

Yazarlar:

Nazmiye Bozok | Emre Akarslan | Said Mahmut Çınar

Yayıncı:

Gumushane University

Ana Başlıklar:

Modelleme ve Simülasyon | Fotovoltaik Güç Sistemleri | Elektrik Mühendisliği (Diğer)

Dergi:

Gümüşhane Üniversitesi Fen Bilimleri Dergisi Vol.15 Issue 1, pp.75-92

Belge Tipi:

Makale

Yayın Tarihi:

3/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001342335

pISSN:

2146-538X

Anahtar kelimeler:

Özet:

 The solar radiation value carries important information about the amount of energy we can obtain from the sun, and predicting the future value of radiation is necessary for proper planning. In this context, various prediction models have been developed, and recently, hybrid models developed to take advantage of the strengths of different models have come to the forefront. This study examines the success of different hybrid model formation strategies in solar radiation prediction. Two regression models AR and RIDGE and two artificial intelligence methods Artificial Neural Network ANN and Elman Neural Network ENN were selected as reference methods. Four different hybrid models were created using different hybrid model formation strategies from these four methods. Solar radiation data were first modeled individually using each method and then the same process was carried out using the created hybrid models. The results obtained with the hybrid models were compared both with each other and with the results obtained from individual models, and their performances were evaluated.

 The solar radiation value carries important information about the amount of energy we can obtain from the sun, and predicting the future value of radiation is necessary for proper planning. In this context, various prediction models have been developed, and recently, hybrid models developed to take advantage of the strengths of different models have come to the forefront. This study examines the success of different hybrid model formation strategies in solar radiation prediction. Two regression models AR and RIDGE and two artificial intelligence methods Artificial Neural Network ANN and Elman Neural Network ENN were selected as reference methods. Four different hybrid models were created using different hybrid model formation strategies from these four methods. Solar radiation data were first modeled individually using each method and then the same process was carried out using the created hybrid models. The results obtained with the hybrid models were compared both with each other and with the results obtained from individual models, and their performances were evaluated.

 The solar radiation value carries important information about the amount of energy we can obtain from the sun, and predicting the future value of radiation is necessary for proper planning. In this context, various prediction models have been developed, and recently, hybrid models developed to take advantage of the strengths of different models have come to the forefront. This study examines the success of different hybrid model formation strategies in solar radiation prediction. Two regression models AR and RIDGE and two artificial intelligence methods Artificial Neural Network ANN and Elman Neural Network ENN were selected as reference methods. Four different hybrid models were created using different hybrid model formation strategies from these four methods. Solar radiation data were first modeled individually using each method and then the same process was carried out using the created hybrid models. The results obtained with the hybrid models were compared both with each other and with the results obtained from individual models, and their performances were evaluated.

 Güneş ışınım değeri, güneşten sağlayabileceğimiz enerji miktarı konusunda önemli bir bilgi taşır ve uygun bir planlama için ışınımın gelecek değerinin kestirilebilmesi gerekir. Bu kapsamda çok farklı tahmin modelleri geliştirilmiş olup, son dönemde, farklı modellerin üstünlüklerinden faydalanmak amacıyla geliştirilen hibrit modeller ön plana çıkmaktadır. Bu çalışmada farklı hibrit model oluşturma stratejilerinin güneş ışınım tahminindeki başarısı incelenmiştir. Bu kapsamda iki regresyon modeli AR ve RIDGE ile iki yapay zekâ yöntemi Yapay Sinir Ağı YSA ve Elman Sinir Ağı ENN referans yöntemler olarak seçilmiştir. Bu dört yöntemden farklı hibrit model oluşturma stratejileri kullanılarak 4 farklı hibrit model oluşturulmuştur. Güneş ışınım verileri önce her bir yöntem bireysel kullanılarak modellenmiş sonra aynı süreç oluşturulan hibrit modeller ile gerçekleştirilmiştir. Hibrit modeller ile elde edilen sonuçlar hem birbirleri ile hem de bireysel modellerden elde edilen sonuçlar ile karşılaştırılmış ve performansları değerlendirilmiştir.

Makale detayları

Comparison of successes of different hybrid method formation strategies in short-term solar radiation prediction

Yazarlar:

Nazmiye Bozok | Emre Akarslan | Said Mahmut Çınar

Yayıncı:

Gumushane University

Ana Başlıklar:

Modelleme ve Simülasyon | Fotovoltaik Güç Sistemleri | Elektrik Mühendisliği (Diğer)

Dergi:

Gümüşhane Üniversitesi Fen Bilimleri Dergisi Vol.15 Issue 1, pp.75-92

Belge Tipi:

Makale

Yayın Tarihi:

3/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001342335

pISSN:

2146-538X

Solar radiation

Hybrid strategy

Regression model

Artificial neural network modelGüneş ışınımı

Hibrit strateji

Regresyon modeli

Yapay sinir ağı modeli

Sayı

Makale detayları

Yazarlar: Nazmiye Bozok | Emre Akarslan | Said Mahmut Çınar

Yayıncı: Gumushane University

Ana Başlıklar: Modelleme ve Simülasyon | Fotovoltaik Güç Sistemleri | Elektrik Mühendisliği (Diğer)

Dergi: Gümüşhane Üniversitesi Fen Bilimleri Dergisi Vol.15 Issue 1, pp.75-92

Belge Tipi: Makale

Yayın Tarihi: 3/2025

Belge Dili: Türkçe

Document ID (Org No.): 2001342335

pISSN: 2146-538X

Sayı

Yazarlar:

Nazmiye Bozok | Emre Akarslan | Said Mahmut Çınar

Yayıncı:

Gumushane University

Ana Başlıklar:

Modelleme ve Simülasyon | Fotovoltaik Güç Sistemleri | Elektrik Mühendisliği (Diğer)

Dergi:

Gümüşhane Üniversitesi Fen Bilimleri Dergisi Vol.15 Issue 1, pp.75-92

Belge Tipi:

Makale

Yayın Tarihi:

3/2025

Belge Dili:

Türkçe

Document ID (Org No.):

2001342335

pISSN:

2146-538X