Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM68862
Volume:
10 |
11 |
Nov 2026
Pages:
1-4
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.

Department of Information Technology
Prof. Ram Meghe College of Engineering & Management, Badnera–Amravati, Maharashtra, India

Department of Information Technology
Prof. Ram Meghe College of Engineering & Management, Badnera–Amravati, Maharashtra, India

B.E. (Information Technology), Department of Information Technology
Prof. Ram Meghe College of Engineering & Management, Badnera–Amravati, Maharashtra, India

B.E. (Information Technology), Department of Information Technology
Prof. Ram Meghe College of Engineering & Management, Badnera–Amravati, Maharashtra, India

B.E. (Information Technology), Department of Information Technology
Prof. Ram Meghe College of Engineering & Management, Badnera–Amravati, Maharashtra, India

B.E. (Information Technology), Department of Information Technology
Prof. Ram Meghe College of Engineering & Management, Badnera–Amravati, Maharashtra, India
Solar photovoltaic (PV) installations are increasingly deployed across residential, commercial and utility-scale environments. However, panel surfaces are continuously exposed to dust accumulation, shading, cracks and impact-related physical damage, all of which can reduce effective energy generation and increase maintenance requirements. Conventional inspection relies heavily on periodic visual checks and electrical testing, which can be labour-intensive and difficult to scale across large installations. This paper presents a proposed real-time, image-based solar-panel fault detection and maintenance recommendation framework using deep learning. The approach accepts RGB images captured by an inspector, camera or drone; applies image preprocessing; and uses convolutional neural-network transfer-learning architectures, particularly VGG16 and MobileNetV2, to classify visible panel conditions. The predicted class is passed to a severity-assessment and rule-based recommendation module that maps conditions such as dust, shading, cracks and physical damage to practical maintenance actions. A React-based web interface is proposed to display the analyzed image, predicted category, confidence score, severity level, recommendation and scan history. The paper reviews related AI-based PV diagnosis research, identifies gaps in real-time multi-fault detection and diagnosis-to-action integration, and describes the proposed system architecture, workflow, technologies, benefits, limitations and future enhancements. As the work is presented as a proposed framework, model performance metrics are reserved for implementation and validation rather than reported as completed experimental results.
KEYWORDS
This paper presents a detailed design for a deep-learning-based solar-panel fault detection and recommendation framework. The proposed workflow processes RGB panel images, applies preprocessing and CNN-based feature extraction, and uses VGG16 or MobileNetV2 transfer learning to classify visible conditions such as normal, dust, cracks, physical damage and shading. A subsequent severity-assessment stage and rule-based recommendation engine translate the predicted condition into a practical maintenance action. A React-based web interface is proposed to display the image, class prediction, confidence, severity and recommendation. The approach responds to identified gaps in multi-fault image classification, efficient inference and integration of diagnosis with maintenance guidance. Future implementation should validate the dataset, model performance, severity rules and field usability before making quantitative performance claims or deploying the system operationally.
[1] Gokulan S., Rajan C., Lalith Kishore V. S., and Suthi G., “Automated Solar Panel Fault Classification Using VGG16-Based Transfer Learning,” IEEE ICISS, 2026. [2] Aditi Chatterjee, Sudhanshu Prakash, Subham Sharma, Debasish Panda, and Nabadeep Patra, “A Hybrid Machine Learning Model for Smart Monitoring and Predictive Maintenance of Solar Panels,” IEEE ICEESI, 2026. [3] Vivek Tetarwal, “A Hybrid Ensemble Learning Framework for Image-Based Solar Panel Classification,” IEEE GCON, June 2025. [4] Indri Windriasari, Aradea, Rianto, and Rina Mardiati, “Computer Vision for Smart Energy Systems: A Review on Deep Learning Approaches for Solar Panel Fault Detection,” IEEE, 2025. [5] Sarmad Rostami et al., “Convolutional Neural Networks for Hotspot Detection in Photovoltaic Systems,” IEEE ICECET, July 2025. [6] Geetha R., “Efficient Fault Detection in Solar Panels Using MobileNetV2-Based Deep Learning Architecture,” IEEE ICECA, 2025. [7] Salaki Reynaldo Joshua, Sanguk Park, and Kihyeon Kwon, “EfficientNetB0 for AI-Powered Solar Panel Maintenance: Advanced Fault Detection,” IEEE ICTC, October 2024. [8] Rohit Saravana, Sanjay Aaditya, and Nikkath Bushra, “Fault Detection in Solar Panels Using Time Series Analysis and CNN,” IEEE ICoACT, 2025. [9] Dilantha Haputhanthri et al., “Generative AI Agents for Hyper Predictive Maintenance of Solar Energy Systems,” 2025. [10] Pompy Barhoi, Himanshu Ladhiya, and Rajib Malik, “Image-Based Fault Detection System for Solar Panels Using Machine Learning,” November 2025.
Publication Date
11/10/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
0
Total views since publication
PDF Downloads
2
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...