Evaluation of Self-Potential Anomalies  caused by Sphere Shaped Structures with Radial Basis Function Neural Networks (RBFNN) and Levenberg-Marquardt(LM) Inversion Methods

Petek Sındırgı; İlknur Kaftan

Makale detayları

Evaluation of Self-Potential Anomalies caused by Sphere Shaped Structures with Radial Basis Function Neural Networks (RBFNN) and Levenberg-Marquardt(LM) Inversion Methods

Yazarlar:

Petek Sındırgı | İlknur Kaftan

Yayıncı:

Dokuz Eylul University

Ana Başlıklar:

Mühendislik

Dergi:

Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi Vol.25 Issue 73, pp.159-166

Belge Tipi:

Makale

Yayın Tarihi:

1/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001281972

pISSN:

1302-9304

Anahtar kelimeler:

Özet:

 The main purpose of the method is to determine the source structure properties in self-potential SP applications. Various methods are used to determine these properties. In this study, Radial Basis Function Neural Network RBFANN and traditional Levenberg-Marquardt LM inversion methods were applied to SP data. The study consists of two stages. In the first stage, the model parameters were determined by performing inverse solutions of the SP anomaly of the noise-free and noisy spherical polarized synthetic model with both methods. In the second stage, the methods were applied to the SP data collected from the Seferihisar İzmir field. When the results were compared, it was seen that RBFANN determined the model parameters with a relatively smaller error value than the LM inversion results. In conclusion, this study revealed that the SP sphere model inversion can be reliably performed using the RBFANN method.

 The main purpose of the method is to determine the source structure properties in self-potential SP applications. Various methods are used to determine these properties. In this study, Radial Basis Function Neural Network RBFANN and traditional Levenberg-Marquardt LM inversion methods were applied to SP data. The study consists of two stages. In the first stage, the model parameters were determined by performing inverse solutions of the SP anomaly of the noise-free and noisy spherical polarized synthetic model with both methods. In the second stage, the methods were applied to the SP data collected from the Seferihisar İzmir field. When the results were compared, it was seen that RBFANN determined the model parameters with a relatively smaller error value than the LM inversion results. In conclusion, this study revealed that the SP sphere model inversion can be reliably performed using the RBFANN method.

 The main purpose of the method is to determine the source structure properties in self-potential SP applications. Various methods are used to determine these properties. In this study, Radial Basis Function Neural Network RBFANN and traditional Levenberg-Marquardt LM inversion methods were applied to SP data. The study consists of two stages. In the first stage, the model parameters were determined by performing inverse solutions of the SP anomaly of the noise-free and noisy spherical polarized synthetic model with both methods. In the second stage, the methods were applied to the SP data collected from the Seferihisar İzmir field. When the results were compared, it was seen that RBFANN determined the model parameters with a relatively smaller error value than the LM inversion results. In conclusion, this study revealed that the SP sphere model inversion can be reliably performed using the RBFANN method.

 Doğal uçlaşama DU uygulamalarında kaynak yapı özelliklerini belirlemek yöntemin temel amacıdır. Çeşitli yöntemler bu özelliklerin saptanmasında kullanlmaktadır. Bu çalışmada Radyal Taban Fonksiyonlu Yapay Sinir Ağları RTFA ve geleneksel Levenberg-Marquardt LM ters çözüm yöntemleri DU verilerine uygulanmıştır. Çalışma iki aşamadan oluşmaktadır. İlk aşamada, gürültüsüz ve gürültülü küre şekilli kuramsal modelin DU anomalisinin her iki yöntemle ters çözümleri yapılarak model parametreleri saptanmıştır. İkinci aşamada ise yöntemler Seferihisarİzmir alanından toplanmış olan DU verilerine uygulanmıştır. Elde edilen sonuçlar karşılaştırıldığında, RTFAnın LM ters çözüm sonuçlarına göre nispeten daha küçük hata değeriyle model parametrelerini saptadığı görülmüştür. Sonuç olarak, bu çalışma, RTFA yöntemi kullanılarak DU küre modeli ters çözümünün güvenilir bir şekilde yapılabileceğini ortaya koymuştur.

Makale detayları

Evaluation of Self-Potential Anomalies caused by Sphere Shaped Structures with Radial Basis Function Neural Networks (RBFNN) and Levenberg-Marquardt(LM) Inversion Methods

Yazarlar:

Petek Sındırgı | İlknur Kaftan

Yayıncı:

Dokuz Eylul University

Ana Başlıklar:

Mühendislik

Dergi:

Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi Vol.25 Issue 73, pp.159-166

Belge Tipi:

Makale

Yayın Tarihi:

1/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001281972

pISSN:

1302-9304

Self Potential

Radial Basis Function Neural Network (RBFANN)

Levenberg-Marquardt(LM) inversion

Doğal Uçlaşma(DU)

Radyal Taban Fonksiyonlu Yapay Sinir Ağlar (RTFA)

Levenberg-Marquardt(LM) ters çözümü

Sayı

Makale detayları

Evaluation of Self-Potential Anomalies caused by Sphere Shaped Structures with Radial Basis Function Neural Networks (RBFNN) and Levenberg-Marquardt(LM) Inversion Methods

Yazarlar: Petek Sındırgı | İlknur Kaftan

Yayıncı: Dokuz Eylul University

Ana Başlıklar: Mühendislik

Dergi: Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi Vol.25 Issue 73, pp.159-166

Belge Tipi: Makale

Yayın Tarihi: 1/2023

Belge Dili: Türkçe

Document ID (Org No.): 2001281972

pISSN: 1302-9304

Self Potential

Radial Basis Function Neural Network (RBFANN)

Levenberg-Marquardt(LM) inversion

Doğal Uçlaşma(DU)

Radyal Taban Fonksiyonlu Yapay Sinir Ağlar (RTFA)

Levenberg-Marquardt(LM) ters çözümü

Sayı

Yazarlar:

Petek Sındırgı | İlknur Kaftan

Yayıncı:

Dokuz Eylul University

Ana Başlıklar:

Mühendislik

Dergi:

Dokuz Eylül Üniversitesi Mühendislik Fakültesi Fen ve Mühendislik Dergisi Vol.25 Issue 73, pp.159-166

Belge Tipi:

Makale

Yayın Tarihi:

1/2023

Belge Dili:

Türkçe

Document ID (Org No.):

2001281972

pISSN:

1302-9304