A New Forecasting Approach for Wind Energy Efficiency with Artificial Neural Networks: The Case of Bozcaada in Çanakkale Province

Adem Demirtop; Ali Hakan Işık

Makale detayları

A New Forecasting Approach for Wind Energy Efficiency with Artificial Neural Networks: The Case of Bozcaada in Çanakkale Province

Yazarlar:

Adem Demirtop | Ali Hakan Işık

Yayıncı:

Burdur Mehmet Akif Ersoy University

Ana Başlıklar:

Rüzgar Enerjisi Sistemleri

Dergi:

Uluslararası Mühendislik Tasarım ve Teknoloji Dergisi Vol.5 Issue 2-Jan, pp.25-32

Belge Tipi:

Makale

Yayın Tarihi:

12/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001309802

pISSN:

2667-5374

Anahtar kelimeler:

Özet:

 In this study, a new efficiency approach is proposed with artificial neural networks ANN and matrix laboratory MATLAB program, which is the program to be applied, to determine how wind energy, which has a serious investment among the increasing renewable energy sources in the world, can be obtained more efficiently. With this proposed approach, Weka Waikato Environment for Knowledge Analysis developed in Java, which offers a combination of machine learning algorithms and requirements such as data preprocessing, was also used. The data used belong to Bozcaada district of Çanakkale province. In 2014, it includes wind speed, humidity, pressure and temperature data obtained from the General Directorate of Meteorology. Wind speed prediction was made with these data sets. Looking at the wind speed estimation results; In WEKA, using the Linear Regression algorithm, the high error rate of the Cross Correlation CC was 0.6797 and the Root Mean Square Logarithmic Error RMSE was 0.1558. In Matlab, the Levenberg-Marquardt training algorithm was used and the regression value for the training data was R0.9998, the regression value for the validation data was R0.99991, the regression value for the test data was R0.62945 and the regression value for all values data was R0.8971. Our RMSE value was found to be 8.0266e4. From this result, considering the ANN applied for both programs, MATLAB is more accurate than the WEKA program.

 In this study, a new efficiency approach is proposed with artificial neural networks ANN and matrix laboratory MATLAB program, which is the program to be applied, to determine how wind energy, which has a serious investment among the increasing renewable energy sources in the world, can be obtained more efficiently. With this proposed approach, Weka Waikato Environment for Knowledge Analysis developed in Java, which offers a combination of machine learning algorithms and requirements such as data preprocessing, was also used. The data used belong to Bozcaada district of Çanakkale province. In 2014, it includes wind speed, humidity, pressure and temperature data obtained from the General Directorate of Meteorology. Wind speed prediction was made with these data sets. Looking at the wind speed estimation results; In WEKA, using the Linear Regression algorithm, the high error rate of the Cross Correlation CC was 0.6797 and the Root Mean Square Logarithmic Error RMSE was 0.1558. In Matlab, the Levenberg-Marquardt training algorithm was used and the regression value for the training data was R0.9998, the regression value for the validation data was R0.99991, the regression value for the test data was R0.62945 and the regression value for all values data was R0.8971. Our RMSE value was found to be 8.0266e4. From this result, considering the ANN applied for both programs, MATLAB is more accurate than the WEKA program.

 In this study, a new efficiency approach is proposed with artificial neural networks ANN and matrix laboratory MATLAB program, which is the program to be applied, to determine how wind energy, which has a serious investment among the increasing renewable energy sources in the world, can be obtained more efficiently. With this proposed approach, Weka Waikato Environment for Knowledge Analysis developed in Java, which offers a combination of machine learning algorithms and requirements such as data preprocessing, was also used. The data used belong to Bozcaada district of Çanakkale province. In 2014, it includes wind speed, humidity, pressure and temperature data obtained from the General Directorate of Meteorology. Wind speed prediction was made with these data sets. Looking at the wind speed estimation results; In WEKA, using the Linear Regression algorithm, the high error rate of the Cross Correlation CC was 0.6797 and the Root Mean Square Logarithmic Error RMSE was 0.1558. In Matlab, the Levenberg-Marquardt training algorithm was used and the regression value for the training data was R0.9998, the regression value for the validation data was R0.99991, the regression value for the test data was R0.62945 and the regression value for all values data was R0.8971. Our RMSE value was found to be 8.0266e4. From this result, considering the ANN applied for both programs, MATLAB is more accurate than the WEKA program.

 Bu çalışmada, dünya üzerinde giderek artan yenilenebilir enerji kaynakları arasında ciddi bir yatırıma sahip olan rüzgâr enerjisinin daha verimli nasıl elde edilebilirliğini yapay sinir ağlar YSA ve uygulanacak program olan matriks laboratuvar MATLAB programı ile yeni bir verimlilik yaklaşımı önerilmiştir. Önerilen bu yaklaşım ile makine öğrenmesi algoritmalarının ve veri ön işleme gibi gereksinimleri bir arada sunan Java ile geliştirilen Weka Waikato Environment for Knowledge Analysis da kullanılmıştır. Kullanılan veriler Çanakkale ili Bozcaada ilçesine aittir. 2014 yılında Meteoroloji Genel Müdürlüğünden temin edilen rüzgâr hızı, nem, basınç ve sıcaklık verilerini kapsamaktadır. Bu veri setleriyle rüzgâr hız tahmin yapılmıştır. Rüzgâr hızı tahmin sonuçlarına bakıldığında; WEKA da Lineer Regresyon algoritmasından faydalanarak korelasyon katsayısının yüksek hata oranını Cross Correlation, CC 0.6797 ve Kök Ortalama Hata Karesi Root Mean Square Logaritmic Error, RMSE 0.1558 değeri düşük çıkmıştır. Matlabda ise Levenberg-Marquardt eğitim algoritması kullanılmıştır ve eğitim verisi için regresyon değeri R0.9998, doğrulama verisi için regresyon değeri R0,99991, test verisi regresyon değer R0,62945 ve tüm değerler verisi için regresyon olup değeri R0,8971 olarak elde edilmiştir. RMSE değerimiz ise 8,0266e4 olarak bir değer bulunmuştur. Buradan sonuçla girilen verililerimiz her iki program için uygulanan YSA göz önünde bulundurularak MATLAB, WEKA programına göre daha iyi sonuçlar elde ettiği gözlemlenmiştir.

Makale detayları

A New Forecasting Approach for Wind Energy Efficiency with Artificial Neural Networks: The Case of Bozcaada in Çanakkale Province

Yazarlar:

Adem Demirtop | Ali Hakan Işık

Yayıncı:

Burdur Mehmet Akif Ersoy University

Ana Başlıklar:

Rüzgar Enerjisi Sistemleri

Dergi:

Uluslararası Mühendislik Tasarım ve Teknoloji Dergisi Vol.5 Issue 2-Jan, pp.25-32

Belge Tipi:

Makale

Yayın Tarihi:

12/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001309802

pISSN:

2667-5374

Renewable energy sources

Wind energy

Artificial neural networks

MATLAB

WEKAYenilenebilir enerji kaynakları

Rüzgâr enerjisi

Yapay sinir ağları

MATLAB

WEKA

Sayı

Makale detayları

Yazarlar: Adem Demirtop | Ali Hakan Işık

Yayıncı: Burdur Mehmet Akif Ersoy University

Ana Başlıklar: Rüzgar Enerjisi Sistemleri

Dergi: Uluslararası Mühendislik Tasarım ve Teknoloji Dergisi Vol.5 Issue 2-Jan, pp.25-32

Belge Tipi: Makale

Yayın Tarihi: 12/2023

Belge Dili: Türkçe

Document ID (Org No.): 2001309802

pISSN: 2667-5374

Sayı

Yazarlar:

Adem Demirtop | Ali Hakan Işık

Yayıncı:

Burdur Mehmet Akif Ersoy University

Ana Başlıklar:

Rüzgar Enerjisi Sistemleri

Dergi:

Uluslararası Mühendislik Tasarım ve Teknoloji Dergisi Vol.5 Issue 2-Jan, pp.25-32

Belge Tipi:

Makale

Yayın Tarihi:

12/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001309802

pISSN:

2667-5374