Fotovoltaik (PV) Sistemlerde Anlık Hata Tespitinin Belirlemesinde Bir Boyutlu Konvolüsiyonel Sinir Ağı (CNN) Kullanımı
Loading...

Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Open Access Color
OpenAIRE Downloads
OpenAIRE Views
Abstract
Bu çalışmada, şebekeye bağlı fotovoltaik (PV) sistemlerde 1D CNN modeli kullanılarak arıza tespiti belirlenmesi sağlanmıştır. Artan enerji ihtiyacı ve yenilenebilir enerji kaynaklarına olan ilginin bir sonucu olarak, PV sistemlerin güvenilirliği ve verimliliği kritik önem taşımaktadır. PV sistemlerde arızaların erken tespiti, enerji üretiminde sürekliliği sağlamak, enerji kayıplarını önlemek ve bakım maliyetlerini azaltmak açısından büyük bir öneme sahiptir. 1D CNN modeli, doğruluk, üzerinden değerlendirilmiştir. Elde edilen sonuçlar, modelin SVM ve KNN gibi geleneksel yöntemlere kıyasla daha yüksek doğruluk oranları sağladığını ortaya koymuştur. Bu da modelin PV sistemlerde arıza tespiti için güçlü bir alternatif olduğunu göstermektedir. Çalışmada, GPVS-Arızaveri seti kullanılarak PV sistemlerdeki Ipv, Vpv, Vdc, ia, ib, ic, va, vb, vc, Iabc, If, Vabc ve Vf gibi kritik değişkenler incelenmiştir. Bu değişkenler üzerinden grafiksel analizler yapılmış, arızaların neden olduğu sapmalar ve düzensizlikler tespit edilmiştir. Arıza tespiti ve sınıflandırma süreci üç ana adımdan oluşmaktadır. İlk olarak, veriler üzerinde yapılan analizlerle F1, F2, F3 ve F4 dosyalarındaki arıza belirtileri açıkça gözlemlenmiştir. Ardından, tespit edilen arızalar etiketlenerek sınıflandırma için uygun hale getirilmiştir. Anahtar Kelimeler: Şebekeye Bağlı PV Sistemler, 1D CNN (1 Boyutlu Konvolüsyonel Sinir Ağı), Arıza Tespiti
In this study, fault detection in grid-connected photovoltaic (PV) systems was achieved using a one-dimensional convolutional neural network (1D CNN) model. The primary aim of this study is to develop a method that accurately and promptly detects and classifies faults that may occur in grid-connected PV systems. With the increasing demand for energy and growing interest in renewable energy sources, the reliability and efficiency of PV systems have become critically important. Early detection of faults in PV systems is vital for ensuring continuity in energy production, preventing energy losses, and reducing maintenance costs. The performance of the 1D CNN model was evaluated based on its accuracy. The results revealed that the model achieved higher accuracy rates compared to traditional methods like Support Vector Machines (SVM) and K-Nearest Neighbors (KNN). This demonstrates that the 1D CNN model is a robust alternative for fault detection in PV systems The GPVS-Fault dataset was used in the study, which analyzed the critical variables in PV systems such as Ipv, Vpv, Vdc, ia, ib, ic, va, vb, vc, Iabc, If, Vabc and Vf. The graphical analysis of these variables highlighted the deviations and irregularities caused by the faults. The fault detection and classification process consisted of three main steps. First, through data analysis, fault indicators were clearly observed in the F1, F2, F3 and F4 files. Then, the detected faults were labeled and prepared for classification. Keywords: Grid-Connected PV Systems, 1D CNN, Fault Detection
In this study, fault detection in grid-connected photovoltaic (PV) systems was achieved using a one-dimensional convolutional neural network (1D CNN) model. The primary aim of this study is to develop a method that accurately and promptly detects and classifies faults that may occur in grid-connected PV systems. With the increasing demand for energy and growing interest in renewable energy sources, the reliability and efficiency of PV systems have become critically important. Early detection of faults in PV systems is vital for ensuring continuity in energy production, preventing energy losses, and reducing maintenance costs. The performance of the 1D CNN model was evaluated based on its accuracy. The results revealed that the model achieved higher accuracy rates compared to traditional methods like Support Vector Machines (SVM) and K-Nearest Neighbors (KNN). This demonstrates that the 1D CNN model is a robust alternative for fault detection in PV systems The GPVS-Fault dataset was used in the study, which analyzed the critical variables in PV systems such as Ipv, Vpv, Vdc, ia, ib, ic, va, vb, vc, Iabc, If, Vabc and Vf. The graphical analysis of these variables highlighted the deviations and irregularities caused by the faults. The fault detection and classification process consisted of three main steps. First, through data analysis, fault indicators were clearly observed in the F1, F2, F3 and F4 files. Then, the detected faults were labeled and prepared for classification. Keywords: Grid-Connected PV Systems, 1D CNN, Fault Detection
Description
Keywords
Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
Turkish CoHE Thesis Center URL
Fields of Science
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
End Page
87
