TK1102 : Fault detection and classificationin associated with photovoltaic arrays within DC microgrids using deep learning approach
Thesis > Central Library of Shahrood University > Electrical Engineering > MSc > 2026
Authors:
[Author], Mojtaba Shivaie[Supervisor]
Abstarct: Abstract With the rapid growth of renewable energy use, photovoltaic systems have gained a prominent position as one of the most important sources of clean electricity generation. However, the occurrence of faults in these systems can lead to reduced efficiency, increased maintenance costs, and disruption in sustainable operation. Therefore, accurate fault detection and classification using intelligent methods play a key role in improving efficiency and extending the lifespan of photovoltaic systems. This thesis focuses on fault detection and classification in grid-connected photovoltaic distribution networks using machine learning and deep learning algorithms. The main objective is to compare the performance of these approaches in fault detection across three different configurations of photovoltaic systems: grid-connected, off-grid, and hybrid. For this purpose, real-world data on solar radiation and temperature from different seasons of the year were used to simulate the systems. In this thesis, in addition to using statistical features, spectral analysis techniques were employed to extract features and improve the accuracy of
Keywords:
#Feb 2026 Abstract With the rapid growth of renewable energy use #photovoltaic systems have gained a prominent position as one of the most important sources of clean electricity generation. However #the occurrence of faults in these systems can lead to reduced efficiency #increased maintenance costs #and disruption in sustainable operation. Therefore #accurate fault detection and classification using intelligent methods play a key role in improving efficiency and extending the lifespan of photovoltaic systems. This thesis focuses on fault detection and classification in grid-connected photovoltaic distribution networks using machine learning and deep learning algorithms. The main objective is to compare the performance of these approaches in fault detection across three different configurations of photovoltaic systems: grid-connected #off-grid #and hybrid. For this purpose #real-world data on solar radiation and temperature from different seasons of the year were used to simulate the systems. In this thesis #in addition to using statistical features #spectral analysis techniques were employed to extract features and improve the accuracy of the models. The results showed that machine learning algorithms such as LightGBM #Random Forest #Extra Trees #AdaBoost #CatBoost #XGBoost #MLP #and SVM outperformed deep learning models like CNN #ResNet #TCN #Transformer #and LSTM in fault detection. Machine learning models such as XGBoost and Random Forest perform relatively better in grid-connected systems with accuracies of approximately 97% and 97.7% #respectively #compared to deep learning models like CNN and ResNet19 #which have accuracies of 95.9% and 95.6% #respectively. These results indicate that the choice of the appropriate method depends on the type of data and the complexity of the system Keywords: Deep Learning #Fault Detection and Classification #Machine Learning #Photovoltaic Systems #SVM machine learning algorithm. International Journal of Intelligent Systems and Applications in Engineering # Keeping place: Central Library of Shahrood University
Visitor: