Mostafa A. Ebied
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Mostafa A. Ebied has a total of 22 co-authors. Here are the top co-authors by number of shared publications:
Amr Munshi
Department of Computer and Network Engineering, College of Computing, Umm Al-Qura University, Makkah, Saudi Arabia
2 shared publications- Abdelhalim Zekry
Department of Electronics and Communications, Faculty of Engineering, Ain Shams University, Cairo, Egypt
1 shared publication - Adel Shaker El-Fishawy
Menoufia University
1 shared publication - Ahmed Sedik
Smart Systems Engineering Laboratory, College of Engineering, Prince Sultan University, Riyadh, Saudi Arabia
1 shared publication Amr I. Shehta
Department of Information System, Faculty of Computers and Artificial Intelligence, Helwan University, Cairo, Egypt
1 shared publication- Ashraf A. M. Khalaf
Department of Electronics and Electrical Communications Engineering, Minia University, Minia, Egypt
1 shared publication
Latest Publications
- Top Cited PapersOpen Access
- 27 August 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-24
Advanced deep learning modeling to enhance detection of defective photovoltaic cells in electroluminescence images
This paper discusses a deep learning approach for detecting defects in photovoltaic (PV) modules using electroluminescence (EL) images. The method addresses key challenges in two practical areas: Creating high-quality EL images to overcome imbalance issues in existing datasets. This is accomplished by employing generative adversarial network (GAN) properties to generate new images. Enhancing training efficiency and performance through a one-cycle policy with optimized learning rate settings, designed to overcome hardware limitations. The research highlights that while automatic defect classification in PV modules is gaining attention as an alternative to visual/manual inspection, the process remains challenging due to the inhomogeneous nature of cell cracks and complex backgrounds in crystalline solar cells. A comparison was made between popular deep learning models (Densenet169, Densenet201, Resnet101, Resnet152, Senet154, Vgg16, and Vgg19) to assess the effectiveness of our approaches on multiple variants of our dataset. We also observe a shift in the phenomenon of moving the threshold in regression estimates because of employing a policy that uses a dynamic threshold instead of a standard threshold (0.5). We have employed two different categorizations that use binary numbers; the first employs four classes (0%, 33%, 67%, and 100%), while the second employs eight classes that are identical to four classes. However, each class has two varieties (monocrystalline and polycrystalline) and a boundary beyond which results will be obtained. Based on the performance results, it was found that the pre-trained Resnet152 model achieved the highest classification accuracy (90.13% for Datasets) of all approaches. Additionally, we have demonstrated that approaches that utilize over-sampling have the greatest performance. These findings emphasize the strength and innovation of our approach, combining advanced data augmentation, adaptive thresholding, and optimized learning strategies. The proposed system not only achieved a peak classification accuracy of 90.13% using ResNet152 but also demonstrated high robustness, reduced training time, and superior generalization across defect types and cell categories. This positions our framework as a scalable and deployment-ready solution for real-world photovoltaic quality inspection systems.
- 27 August 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-24
Top Cited PapersOpen AccessAdvanced deep learning modeling to enhance detection of defective photovoltaic cells in electroluminescence images
This paper discusses a deep learning approach for detecting defects in photovoltaic (PV) modules using electroluminescence (EL) images. The method addresses key challenges in two practical areas: Creating high-quality EL images to overcome imbalance issues in existing datasets. This is accomplished by employing generative adversarial network (GAN) properties to generate new images. Enhancing training efficiency and performance through a one-cycle policy with optimized learning rate settings, designed to overcome hardware limitations. The research highlights that while automatic defect classification in PV modules is gaining attention as an alternative to visual/manual inspection, the process remains challenging due to the inhomogeneous nature of cell cracks and complex backgrounds in crystalline solar cells. A comparison was made between popular deep learning models (Densenet169, Densenet201, Resnet101, Resnet152, Senet154, Vgg16, and Vgg19) to assess the effectiveness of our approaches on multiple variants of our dataset. We also observe a shift in the phenomenon of moving the threshold in regression estimates because of employing a policy that uses a dynamic threshold instead of a standard threshold (0.5). We have employed two different categorizations that use binary numbers; the first employs four classes (0%, 33%, 67%, and 100%), while the second employs eight classes that are identical to four classes. However, each class has two varieties (monocrystalline and polycrystalline) and a boundary beyond which results will be obtained. Based on the performance results, it was found that the pre-trained Resnet152 model achieved the highest classification accuracy (90.13% for Datasets) of all approaches. Additionally, we have demonstrated that approaches that utilize over-sampling have the greatest performance. These findings emphasize the strength and innovation of our approach, combining advanced data augmentation, adaptive thresholding, and optimized learning strategies. The proposed system not only achieved a peak classification accuracy of 90.13% using ResNet152 but also demonstrated high robustness, reduced training time, and superior generalization across defect types and cell categories. This positions our framework as a scalable and deployment-ready solution for real-world photovoltaic quality inspection systems.
- Top Cited PapersOpen Access
- 28 March 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-27
Evaluating machine learning models comprehensively for predicting maximum power from photovoltaic systems
This paper presents a machine learning (ML) model designed to track the maximum power point of standalone Photovoltaic (PV) systems. Due to the nonlinear nature of power generation in PV systems, influenced by fluctuating weather conditions, managing this nonlinear data effectively remains a challenge. As a result, the use of ML techniques to optimize PV systems at their MPP is highly beneficial. To achieve this, the research explores various ML algorithms, such as Linear Regression (LR), Ridge Regression (RR), Lasso Regression (Lasso R), Bayesian Regression (BR), Decision Tree Regression (DTR), Gradient Boosting Regression (GBR), and Artificial Neural Networks (ANN), to predict the MPP of PV systems. The model utilizes data from the PV unit’s technical specifications, allowing the algorithms to forecast maximum power, current, and voltage based on given irradiance and temperature inputs. Predicted data is also used to determine the boost converter’s duty cycle. The simulation was conducted on a 100 kW solar panel with an open-circuit voltage of 64.2 V and a short-circuit current of 5.96 A. Model performance was evaluated using metrics such as Root Mean Square Error (RMSE), Coefficient of Determination (R2), and Mean Absolute Error (MAE). Additionally, the study assessed the correlation and feature importance to evaluate model compatibility and the factors impacting the predictive accuracy of the ML models. Results showed that the DTR algorithm outperformed others like LR, RR, Lasso R, BR, GBR, and ANN in predicting the maximum current (Im), voltage (Vm), and power (Pm) of the PV system. The DTR model achieved RMSE, MAE, and R2 values of 0.006, 0.004, and 0.99999 for Im, 0.015, 0.0036, and 0.99999 for Vm, and 2.36, 0.871, and 0.99999 for Pm. Factors such as the size of the training dataset, operating conditions of the PV system, model type, and data preprocessing were found to significantly influence prediction accuracy.
- 28 March 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-27
Top Cited PapersOpen AccessEvaluating machine learning models comprehensively for predicting maximum power from photovoltaic systems
This paper presents a machine learning (ML) model designed to track the maximum power point of standalone Photovoltaic (PV) systems. Due to the nonlinear nature of power generation in PV systems, influenced by fluctuating weather conditions, managing this nonlinear data effectively remains a challenge. As a result, the use of ML techniques to optimize PV systems at their MPP is highly beneficial. To achieve this, the research explores various ML algorithms, such as Linear Regression (LR), Ridge Regression (RR), Lasso Regression (Lasso R), Bayesian Regression (BR), Decision Tree Regression (DTR), Gradient Boosting Regression (GBR), and Artificial Neural Networks (ANN), to predict the MPP of PV systems. The model utilizes data from the PV unit’s technical specifications, allowing the algorithms to forecast maximum power, current, and voltage based on given irradiance and temperature inputs. Predicted data is also used to determine the boost converter’s duty cycle. The simulation was conducted on a 100 kW solar panel with an open-circuit voltage of 64.2 V and a short-circuit current of 5.96 A. Model performance was evaluated using metrics such as Root Mean Square Error (RMSE), Coefficient of Determination (R2), and Mean Absolute Error (MAE). Additionally, the study assessed the correlation and feature importance to evaluate model compatibility and the factors impacting the predictive accuracy of the ML models. Results showed that the DTR algorithm outperformed others like LR, RR, Lasso R, BR, GBR, and ANN in predicting the maximum current (Im), voltage (Vm), and power (Pm) of the PV system. The DTR model achieved RMSE, MAE, and R2 values of 0.006, 0.004, and 0.99999 for Im, 0.015, 0.0036, and 0.99999 for Vm, and 2.36, 0.871, and 0.99999 for Pm. Factors such as the size of the training dataset, operating conditions of the PV system, model type, and data preprocessing were found to significantly influence prediction accuracy.
- 24 July 2023
- Published by Springer Nature in Wireless Personal Communications
- Vol. 131 (4) , 2331-2369
A Proposed Deep-Learning-Based Framework for Medical Image Communication, Storage and Diagnosis
- 24 July 2023
- Published by Springer Nature in Wireless Personal Communications
- Vol. 131 (4) , 2331-2369
A Proposed Deep-Learning-Based Framework for Medical Image Communication, Storage and Diagnosis