ANN-Based robust framework for accurate fault identification in solar PV system
|
Full Text |
Pdf
|
|
Author |
G. Venkata Suresh Reddy, Immanuel Anupalli and P.Sudheer
|
|
e-ISSN |
1819-6608 |
|
On Pages
|
751-760
|
|
Volume No. |
21
|
|
Issue No. |
11
|
|
Issue Date |
August 10, 2026
|
|
DOI |
https://doi.org/10.59018/062680
|
|
Keywords |
solar PV systems, fault detection and classification, machine learning, naïve Bayes, k-nearest neighbors, logistic regression, artificial neural networks.
|
Abstract
Solar photovoltaic (PV) systems require advanced and intelligent fault detection mechanisms to ensure efficient energy generation under varying environmental conditions. This work proposes a deep learning-based fault classification framework using Artificial Neural Networks (ANN) to accurately detect faults in PV systems operating under nonlinear and noisy conditions. The dataset is constructed from electrical characteristics of PV panels, including current-voltage (I-V) and power-voltage (P-V) curves, under three operating conditions: Healthy, Shading, and Open-Circuit faults. Key features such as open-circuit voltage, short-circuit current, maximum power point parameters, fill factor, and statistical indicators are used to represent system behavior. The proposed ANN model is trained using optimized network architecture and hyperparameter tuning to capture complex nonlinear relationships in the data. The performance of the ANN model is evaluated and compared with traditional machine learning techniques such as Logistic Regression, Naïve Bayes, and K-Nearest Neighbors (KNN). Experimental results show that the ANN model outperforms conventional methods, achieving higher classification accuracy (~97.8%), precision, recall, and F1-score, thereby demonstrating superior capability in handling noisy and nonlinear PV data. The proposed framework is efficient, reliable, and suitable for real-time PV monitoring systems. It enhances predictive maintenance, reduces power losses, and improves the overall performance and reliability of solar energy systems.
Back