Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988

A Holistic Approach to Defect Detection in Solar

Modules: Leveraging Lifecycle Data for Improved
Performance
A. Phadke1; N. Bhagwat2; S. Mohaadkar3; V. Dalal4; P.S. Joshi5
5 Assistant Professor
1,2,3,4,5 Information Technology, MMCOE, Savitribai Phule Pune University, Karve Nagar, Pune, India

Publication Date: 2025/04/30

Abstract—The efficient maintenance and optimization of solar modules are critical for sustaining high energy yields over
their operational lifetimes. This research introduces a comprehensive system designed to enhance lifecycle traceability and
defect detection in solar modules using a combination of advanced image analysis and machine learning techniques. By
leveraging Convolutional Neural Networks (CNN), You Only Look Once (YOLO) object detection, and deep learning, the
system analyzes thermal and normal imaging data as well as current-voltage (IV) characteristics and curves. The proposed
framework enables the detection of common faults, such as hotspots, cell cracks, and degradation patterns, which can impact
performance and safety. Integrated data management and tracking capabilities facilitate end-to-end lifecycle monitoring,
providing accessible, organized insights for stakeholders involved in solar module maintenance and diagnostics. The model
shows an accuracy of 90%. The results show that the system not only improves accuracy in fault identification but also
allows efficient storage and retrieval of diagnostic data, presenting a robust solution for advancing photovoltaic asset
management

How to Cite: A. Phadke; N. Bhagwat; S. Mohaadkar; V. Dalal; P.S. Joshi (2025), A Holistic Approach to Defect Detection in Solar
Modules: Leveraging Lifecycle Data for Improved Performance. International Journal of Innovative Science
and Research Technology, 10(4), 1839-1848. https://doi.org/10.38124/ijisrt/25apr988

I. INTRODUCTION valuable insights across the PV lifecycle.

As global reliance on renewable energy sources Our paper aims to address key challenges in solar
intensifies, photovoltaic (PV) solar panels have become module maintenance, such as the need for real-time, accurate
pivotal in the transition towards sustainable energy. However, fault detection and lifecycle data accessibility, to support
the efficiency of these solar modules can decline due to informed decision-making and predictive maintenance.
environmental stressors, operational issues, and inherent Through this work, we demonstrate the potential of
material degradation over time. Ensuring consistent integrating state-of-the-art image analysis and data
performance and maximizing lifespan thus require effective management within the solar energy sector, driving
and proactive maintenance strategies. Traditional manual improvements in efficiency, reliability, and safety of
inspection methods are often labor-intensive, costly, and may photovoltaic systems.
miss subtle indicators of potential failures, highlighting the
need for automated and precise diagnostic solutions. II. BACKGROUND

This paper proposes a novel system for lifecycle  System Overview

traceability and defect detection in solar modules by The proposed system integrates advanced defect
integrating computer vision and machine learning. detection and data traceability throughout the solar module
Employing Convolutional Neural Networks (CNN) and the lifecycle. It employs deep learning (DL) techniques for
YOLO object detection algorithm, the system processes both accurate and real-time defect identification during
thermal and visible spectrum images to detect faults such as manufacturing and field operations. Data from various stages
hotspots, cracks, and discoloration. Additionally, IV is captured, securely stored, and linked to individual modules
characteristics and their corresponding curves are analyzed to through unique identifiers. DL-based predictive analytics
provide a more holistic view of module health. By embedding enhances fault prediction and maintenance planning. The
deep learning techniques into a structured data management system ensures efficient traceability, quality control, and
framework, the system allows for seamless data tracking, long-term performance monitoring of solar modules. This
storage, and retrieval, thus enabling stakeholders to access comprehensive approach optimizes lifecycle management

IJISRT25APR988 www.ijisrt.com 1839

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988
and operational efficiency. that can occur in solar power setups, including line-to-line
faults, ground faults, and open-circuit issues, and highlights
 Backend Architecture the potential hazards and energy losses associated with
The backend of the solar panel defect detection system undetected faults. The authors discuss both traditional and
ensures seamless coordination of deep learning models and modern diagnostic techniques, ranging from current-voltage
data processing components. It starts with image acquisition (I-V) curve analysis and infrared thermography to advanced
from thermal and visible spectrum cameras, followed by machine learning and artificial intelligence approaches. The
preprocessing to enhance defect detection accuracy. review emphasizes the need for reliable and automated fault
Convolutional Neural Networks (CNNs) and YOLO detection mechanisms to ensure the safety, efficiency, and
algorithms analyze these images to identify defects like longevity of PV systems. The paper also identifies gaps in
hotspots, cracks, and discoloration. Additionally, IV curve existing fault diagnosis technologies and suggests directions
analysis correlates electrical performance with detected faults for future research, making it a valuable resource for those
for a comprehensive assessment. All data is stored in a developing innovative monitoring and protection solutions in
structured database, ensuring easy traceability throughout the the solar energy sector.
module’s lifecycle. A user-friendly interface provides real-
time defect reports and predictive analytics, enabling In this study [2] the authors present a machine vision
proactive maintenance. Automated alerts notify operators of system tailored for detecting edge defects in photovoltaic
critical issues, optimizing efficiency and ensuring long-term (PV) glass. They enhance the SqueezeNet architecture to
solar panel reliability. improve feature extraction capabilities, enabling the accurate
identification of defects such as cracks and chips. The system
 Client Side Interface incorporates high-contrast imaging solutions to capture
The client-side interface of the solar panel defect detailed visuals of the glass edges, which are then processed
detection system is designed for seamless user interaction by the improved SqueezeNet model. Experimental results
with role-based access control for Inspectors, Admins, and demonstrate that this approach achieves high accuracy in
Super Admins. Inspectors can upload images, view detected defect detection while maintaining computational efficiency,
defects, and generate reports based on real-time analysis. making it suitable for real-time industrial applications.
Admins oversee system operations, manage data, and validate
defect reports, ensuring accurate tracking of module health. In this study titled [6],the authors explore the
Super Admins have full control, including user management, integration of thermal imaging and artificial intelligence (AI)
access permissions, and system configuration. The interface to enhance the detection of defects in solar panels. They
provides a dashboard with real-time defect insights, historical propose a system that utilizes infrared cameras to capture
data, and predictive analytics. Secure authentication and thermal images of photovoltaic modules, identifying
access control ensure data integrity while enabling anomalies such as hotspots, cracks, and faulty cells. These
stakeholders to make informed maintenance decisions thermal images are processed using advanced AI algorithms,
efficiently. including convolutional neural networks (CNNs), to
accurately classify and localize defects. The research
III. RELATED WORK demonstrates that combining thermal imaging with AI not
only improves the precision of defect detection but also
Several research works have played an important role enables real-time monitoring, thereby reducing maintenance
in shaping the development of intelligent parking systems. In costs and enhancing the efficiency of solar energy systems.
the paper [10] the authors provide an in-depth review of the The study's findings suggest that this integrated approach is a
protection challenges and fault diagnosis methods in viable solution for automated, efficient, and cost-effective
photovoltaic (PV) systems. It examines various types of faults maintenance of photovoltaic installations.

Table 1 Literature Review

Paper Dataset Objective Methodology Result Challenges
Abdelsattar, Solar cell image Automated defect Deep learning High accuracy in Dataset limitations,
Montaser, et dataset detection in solar algorithms (CNNs) for detecting defects computational
al. (2025) cells image classification complexity
Laguna, G., Minimal data Detecting abnormal Recursive Least Reliable fault Scalability,
et al. (2024) from PV PV system Squares (RLS) detection with adaptability to
systems operations algorithms minimal data different PV
configurations
Rajeshkanna, Real-time solar Fault-related feature Custom deep learning Improved Real-time
R., et al. panel discrimination in network for defect accuracy in defect processing
(2024) manufacturing cell partitioning and identification detection challenges, hardware
data defect classification requirements
Xiong, J., et PV glass edge Defect detection in Improved SqueezeNet- Enhanced Limited dataset,
al. (2024) images PV glass edges based classification detection accuracy edge case handling
with optimized
model

IJISRT25APR988 www.ijisrt.com 1840

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988
Joshua, S. R., Solar-hydrogen Advanced fault Convolutional Neural High fault Integration with
et al. (2024) system data detection in solar- Networks (CNN) detection rate in different PV
hydrogen systems applied to fault hybrid energy systems, false
detection systems positives
Ramadan, E. Photovoltaic Fault detection and Transformer-based Enhanced High computational
A., et al. module data classification using deep learning model accuracy over cost, data
(2024) transformer neural traditional CNNs preprocessing
networks challenges
Umar, S., et Thermal AI-based defect Combination of AI Improved Cost of thermal
al. (2024) imaging data identification in and thermal imaging detection imaging equipment,
solar panels for defect detection efficiency data variability
Giovanardi, IoT-enabled Enabling circular Theoretical IoT Improved Implementation
Matteo, et al. life-cycle data economy via framework for traceability and complexity, data
(2023) lifecycle building façade sustainability security concerns
information flows traceability insights
Abdelsattar, Solar cell image Automated defect Deep learning-based High detection Large-scale
Montaser, et dataset detection in solar image analysis accuracy deployment
al. (2025) cells challenges
Pillai, Various PV Comprehensive Literature review on Identified key Lack of standard
Dhanup S., et system fault review of PV system fault diagnosis challenges and datasets, evolving
al. (2018) datasets protection and fault methods research gaps PV technologies
diagnosis

IV. PROPOSED SYSTEM panel lifecycle. A structured database organizes detected

faults, IV characteristics, and maintenance history, allowing
 System Overview efficient retrieval for predictive analytics and decision-
The proposed solar panel defect detection and data making. Automated alerts notify operators of critical defects,
traceability system leverages deep learning techniques to reducing downtime and optimizing maintenance schedules,
automate fault identification and lifecycle management. It ultimately leading to cost savings and enhanced energy
integrates high-resolution thermal and visible spectrum output.
imaging, coupled with advanced Convolutional Neural
Networks (CNNs) and the YOLO object detection algorithm, The client-side interface is designed with role-based
to detect defects such as hotspots, cracks, and discoloration. access control to facilitate seamless interaction for different
Additionally, IV curve analysis enhances the accuracy of stakeholders. Inspectors can upload images, view defect
fault diagnostics by correlating electrical performance with reports, and generate insights, while Admins manage data
detected defects. This multi-layered approach ensures validation and system operations. Super Admins oversee user
comprehensive monitoring and early fault detection, management, system configurations, and access permissions,
significantly improving the efficiency and reliability of solar ensuring secure and structured data governance. The
panels. interactive dashboard provides real-time insights, historical
trends, and predictive analytics, empowering stakeholders
The backend infrastructure plays a crucial role in with data-driven decision-making for improved solar panel
processing, storing, and managing defect detection data. It maintenance and lifecycle management.
ensures real-time synchronization of module performance
records, enabling seamless tracking throughout the solar  System Architecture

Fig 1 System Architecture

IJISRT25APR988 www.ijisrt.com 1841

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988

Fig 2 Defect Detection Process Flow

 Defect Detection Using Normal Images  Flipping: Ensures defect recognition from multiple
Defect detection using normal (visible spectrum) orientations.
images is a crucial approach in identifying faults in solar  Brightness & Contrast Adjustments: Compensates for
panels. High-resolution cameras capture images under variations in lighting conditions.
natural lighting conditions, enabling the detection of surface-
level defects such as cracks, discoloration, delamination, dirt  Model Training
accumulation, and broken glass. These images are then
processed using computer vision and deep learning  CNN Model Selection:
techniques to enhance fault detection accuracy and automate A Convolutional Neural Network (CNN), VGG16 is
the inspection process. selected for defect detection.

 Data Gathering  Configuring Model Parameters:

Learning rate, batch size, and training epochs are
 Capturing High-Resolution Images: optimized for performance.
Standard cameras are used to capture visible-spectrum
images of solar panels under natural lighting conditions.  Training with Annotated Data:
The labeled dataset is used to train the model, enabling
 Varied Environmental Conditions: it to recognize and classify different surface defects in solar
Images are collected at different times of the day and panels.
under diverse weather conditions to account for variations in
lighting and panel appearances.  Defect Detection

 Image Annotation:  Deploying the CNN Model:

Each image is classified to indicate defects such as The trained model is implemented for real-time analysis
cracks, discoloration, dirt accumulation, delamination, or of incoming images.
broken glass. Proper annotation is essential for training an
accurate machine learning model.  Automated Defect Identification:
The system scans images to detect and classify faults,
 Preprocessing distinguishing between different defect types.

 Image Enhancement: This method offers a cost-effective, non-invasive, and

Techniques such as contrast adjustment, edge detection, scalable solution for monitoring solar panel health. By
and noise reduction improve the clarity of defects, making integrating defect detection with a structured database,
them easier for the model to detect. inspection history can be stored and analyzed over time,
enabling predictive maintenance. While normal image-based
 Data Augmentation: detection is effective for surface-level faults, combining it
Various transformations increase dataset diversity and with infrared thermal imaging and electrical performance
improve the model’s robustness: analysis further enhances the accuracy and reliability of
defect identification capabilities and demonstrates the
 Rotation: Simulates different camera angles and seamless, automated and unattended nature of our smart
perspectives. parking solution.
 Scaling: Represents solar panels at different distances.

IJISRT25APR988 www.ijisrt.com 1842

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988
 Defect Detection Using Thermal Images precise defect identification.

 Data Gathering  Defect Detection

Deploying the YOLO Model: The trained model is
 Capturing Thermal Images: implemented for real-time defect detection. Real-Time
Inspectors use infrared cameras to acquire thermal images of Analysis: Incoming images are analyzed to identify potential
solar panels. defects. Defect Classification: The model categorizes
detected defects based on their type, such as hotspots, cracks,
 Varied Conditions: or faulty cells. Displaying Results: Bounding boxes and
Images are collected at different times and labels are overlaid on images, highlighting defect areas.
environmental conditions to ensure a diverse dataset covering These results are accessible to relevant users, including
multiple panel states. admins and inspectors.

 Image Annotation:  Defect Detection using IV analysis

Each image is labeled to indicate observable defect
types such as hotspots, cracks, or faulty cells. This step is Solar energy adoption is growing exponentially, but
crucial for training the machine learning model to recognize module defects (e.g., microcracks, PID, hot spots) cause 15–
patterns effectively. 25% annual efficiency losses. Traditional defect detection
methods (e.g., electroluminescence, thermal imaging) are
 Preprocessing costly, time-consuming, and lack predictive capabilities.
Enhancing Clarity & Contrast: Image processing
techniques improve visibility and definition of defects,  IV Curve Fundamentals
making them more distinguishable for the model.
 Data Augmentation:  Isc (Short-Circuit Current): Induces shading, soiling, or
Various transformations are applied to increase dataset cell cracks.
diversity and improve model robustness:  Voc (Open-Circuit Voltage): Reflects PID or diode
failures.
 Rotation: Simulates different viewing angles.  Pmax (Maximum Power): Key indicator of overall health.
 Scaling: Represents panels at varying distances.  Fill Factor (FF): Sensitive to series/shunt resistance
 Flipping: Ensures the model can recognize defects in defects.
different orientations.
 AI-Driven Defect Detection Framework
 Model Training Various parameters are measured from portable IV
YOLO Model Selection: A suitable YOLO model tracers such as Pnom, Isc, Voc, M_eff%, FF, Imp, Vmp, etc.
version (e.g., YOLOv9) is chosen for defect detection. The dynamic thresholds for each parameter are adjusted for
Configuring Model Parameters: Key parameters like learning environmental conditions. As an input the IV parameters and
rate, batch size, and detection thresholds are set. environmental data is taken and with help of spatial-temporal
Training with Annotated Data: The preprocessed images, pattern recognition, the output is generated which includes
along with their annotations, are used to train the model for defect type and severity score.

`
Fig 3 Algorithm of Workflow

IJISRT25APR988 www.ijisrt.com 1843

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988
 Use of MongoDB for Storage Additionally, the project incorporates a role-based
MongoDB plays a key role in storing and managing access control system, ensuring secure and structured data
defect detection and lifecycle traceability data for solar governance. Inspectors can upload images and view reports,
panels. As a NoSQL database, it efficiently handles Admins can manage defect validation, and Super Admins
unstructured data like high-resolution images, defect reports, have full system control. The interactive dashboard provides
IV curve data, and maintenance history. Its document-based real-time insights and historical trends, making data-driven
structure allows seamless storage of module-specific details, decision-making more accessible. By integrating deep
including timestamps, location, defect classifications, and learning, real-time defect detection, and intelligent data
performance metrics. management, this project represents a significant
advancement in solar panel monitoring, enhancing reliability,
The system leverages MongoDB’s real-time data efficiency, and sustainability in renewable energy systems.
processing to store defect detection results from deep learning
models, linking each solar panel’s inspection records, faults,  System Configuration
and predictive analytics. Indexing and query optimization The system configuration for solar panel defect
enable fast data retrieval, improving maintenance decisions. detection consists of both hardware and software components
Its horizontal scalability supports large-scale solar farms optimized for efficient deep learning-based analysis. The
without performance loss. hardware setup includes high-resolution infrared and visible
spectrum cameras for capturing detailed images of solar
MongoDB’s role-based access control (RBAC) panels under varying environmental conditions. A GPU-
enhances security by restricting data access for roles like enabled computing system, such as one equipped with
Inspectors and Admins. Real-time synchronization ensures NVIDIA CUDA-compatible GPUs, is essential for
up-to-date information, while replication and backup provide processing deep learning models efficiently. Additionally, a
fault tolerance. Integrating MongoDB ensures efficient cloud-based or on-premise server is used for data storage and
storage, quick retrieval, and seamless lifecycle management, real-time analysis, ensuring seamless accessibility for
boosting defect detection and predictive maintenance. stakeholders.

 Innovation On the software side, the system utilizes Python-based

This project introduces a novel approach to solar panel frameworks, including TensorFlow and PyTorch, for deep
defect detection and lifecycle traceability by integrating deep learning model development. The YOLO object detection
learning, computer vision, and structured data management. algorithm is implemented to detect and classify defects with
Unlike traditional manual inspection methods, which are high accuracy. MongoDB serves as the backend database for
labor-intensive and prone to human error, this system storing inspection data, defect records, and lifecycle
automates fault detection using Convolutional Neural information. A web-based dashboard with role-based access
Networks (CNNs) and the YOLO object detection algorithm. control allows Inspectors, Admins, and Super Admins to
By analyzing both normal (visible spectrum) and thermal interact with the system, view defect reports, and manage
images, the system accurately identifies defects such as solar panel maintenance schedules. The integration of
cracks, discoloration, and hotspots in real-time, significantly automated alerts and predictive analytics enhances decision-
improving inspection efficiency and accuracy. making, making the system highly efficient and scalable.

A key innovation lies in the seamless combination of

defect detection with a robust data traceability framework.
Using MongoDB as the backend database, the system ensures
efficient storage, retrieval, and tracking of inspection records,
IV curve analysis, and maintenance history. Each solar panel
is assigned a unique identifier, allowing stakeholders to
monitor its performance throughout its lifecycle. Predictive
analytics further enhance the system by forecasting potential
failures, enabling proactive maintenance and reducing
operational costs.

IJISRT25APR988 www.ijisrt.com 1844

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988
V. PROOF OF WORK

Fig 4 IV Analysis page(i)

Fig 5 IV-Analysis Page(ii)

IJISRT25APR988 www.ijisrt.com 1845

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988

Fig 6 IV-Analysis Result Page

Fig 7 Visual Image Analysis page(i)

IJISRT25APR988 www.ijisrt.com 1846

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988

Fig 8 Visual Image Analysis page(ii)

Fig 9 Model Implementation

VI. CONCLUSION ensure long-term performance.

This research paper presents an innovative system for Beyond defect detection, the system incorporates
defect detection and data traceability throughout the solar advanced data traceability, enabling seamless tracking of
module lifecycle. By leveraging deep learning for defect module history, performance, and maintenance activities.
identification, the system ensures precise real-time This data-driven approach optimizes manufacturing
monitoring, reducing reliance on manual inspections and processes, enhances lifecycle management, and improves
minimizing operational costs. Automated processes enhance decision-making for stakeholders. By offering automated
efficiency, improving the quality and reliability of solar defect detection, robust traceability, and actionable insights,
modules while enabling predictive maintenance and reducing this system advances intelligent solar module management,
downtime. Manufacturers benefit from increased production contributing to greater efficiency, reliability, and
accuracy, while operators can maximize energy output and sustainability in the solar energy sector.

IJISRT25APR988 www.ijisrt.com 1847

Volume 10, Issue 4, April – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/25apr988
REFERENCES Detection of Solar Panel Defects Using a Thermal
Camera Mobile Application," 2023 10th International
[1] Abdelsattar, Montaser, et al. "Automated Defect Conference on Future Internet of Things and Cloud
Detection in Solar Cell Images Using Deep Learning (FiCloud)
Algorithms." (2025). [15] M. I. Ameerdin, M. H. Jamaluddin, A. Z. Shukor, L.
[2] Laguna, G., Moreno, P., Cipriano, J., Mor, G., Al Hakim Kamaruzaman and S. Mohamad, "Towards
Gabaldón, E., & Luna, A. . Detection of abnormal Efficient Solar Panel Inspection: A YOLO-based
photovoltaic systems’ operation with minimum data Method for Hotspot Detection," 2024
requirements based on Recursive Least Squares [16] S. Lee, K. E. An, B. D. Jeon, K. Y. Cho, S. J. Lee and
algorithms. Solar Energy (2024) D. Seo, "Detecting faulty solar panels based on
[3] Rajeshkanna, R., Meikandan, M., Daxayani, C., & thermal image processing," 2018
Ganesh Kumar, P. Fault-related feature discrimination [17] S. P. Pathak and S. A. Patil, "Evaluation of Effect of
network for cell partitioning and defect classification Pre-Processing Techniques in Solar Panel Fault
in real-time solar panel manufacturing. Proceedings of Detection," in IEEE Access, vol. 11, pp.
the Institution of Mechanical Engineers, Part E: [18] Niazi, Kamran Ali Khan & Akhtar, W. & Khan,
Journal of Process Mechanical Engineering, 238(6), Hassan & Sohaib, Sarmad & Nasir, A.. (2018). Binary
2809-2820. (2024). Classification of Defective Solar PV Modules Using
[4] Xiong, J., He, Z., Zhou, Q., & Yang, R. Photovoltaic Thermography. 10.1109/PVSC.2018.8548138.
glass edge defect detection based on improved [19] Y. Hu, W. Cao, J. Ma, S. J. Finney and D. Li,
SqueezeNet. Signal, Image and Video Processing "Identifying PV Module Mismatch Faults by a
.(2024). Thermography-Based Temperature Distribution
[5] Joshua, S. R., Park, S., & Kwon, K.. Solar Panel Fault Analysis”
Detection: Applying Convolutional Neural Network
for Advanced Fault Detection in Solar-Hydrogen
System at University. In 2024 IEEE 24th International
Conference on Software Quality, Reliability, and
Security Companion (QRS-C) (pp. 289-298).
IEEE.(2024)
[6] Ramadan, E. A., Moawad, N. M., Abouzalm, B. A.,
Sakr, A. A., Abouzaid, W. F., & El-Banby, G. M. . An
innovative transformer neural network for fault
detection and classification for photovoltaic modules.
Energy Conversion and Management (2024)
[7] Umar, S., Qureshi, M. S., & Nawaz, M. U. Thermal
Imaging and AI in Solar Panel Defect Identification.
International Journal of Advanced Engineering
Technologies and Innovations (2024)
[8] Giovanardi, Matteo, et al. "Internet of Things for
building façade traceability: A theoretical framework
to enable circular economy through life-cycle
information flows." Journal of Cleaner Production
382 (2023)
[9] Abdelsattar, Montaser, et al. "Automated Defect
Detection in Solar Cell Images Using Deep Learning
Algorithms." IEEE Access (2025).
[10] Pillai, Dhanup S., and N. Rajasekar. "A
comprehensive review on protection challenges and
fault diagnosis in PV systems." Renewable and
Sustainable Energy Reviews 91 (2018)
[11] S. Dueñas, E. Pérez, H. Castán, H. García and L.
Bailón, "The role of defects in solar cells: Control and
detection defects in solar cells," 2013
[12] Pathak, Sujata & Patil, Sonali. (2023). Evaluation of
Effect of Pre-Processing Techniques in Solar Panel
Fault Detection. IEEE Access. PP. 1-1.
10.1109/ACCESS.2023.3293756.
[13] Gr, Venkatakrishnan & Rengaraj, R. & Tamilselvi, S
& Harshini, J & SahooDetection, location, and
diagnosis of different faults in large solar PV system—
a review.
[14] H. Hajjdiab et al., "Automated Computer Vision-based

IJISRT25APR988 www.ijisrt.com 1848