Industrial Diagnosis of Photovoltaic Systems Using Metaheuristics
No Thumbnail Available
Date
2026
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Photovoltaic (PV) systems are increasingly recognized as effective and sustainable solutions to
meet the global demand for clean energy. However, their efficiency and performance are often
compromised by various technical faults, including partial shading, cell aging, hotspots and
short or open-circuit issues, among other operational challenges. Consequently, the
development of intelligent and interpretable fault diagnosis methods is crucial to ensure the
reliability and sustainability of these systems.
This thesis presents innovative approaches that integrate intelligent optimization techniques and
include metaheuristic algorithms, with explainable artificial intelligence (XAI) to enhance
diagnostic accuracy and modeling efficiency. Initially, a hybrid signal filtering method was
proposed to improve data quality and reduce noise, thereby strengthening the robustness of fault
diagnosis using Grey Wolf Optimization (GWO ) to evaluate algorithmic effectiveness.
Furthermore, an Adaptive variant of GWO (AGWO ) was developed for feature selection,
significantly improving model accuracy and reducing execution time compared to conventional
methods.
To further advance classification capabilities, this research introduces the Discrete Grey Wolf
Optimization (DGWO ) algorithm, which adapts classical GWO for discrete search spaces in
rule-based classification. The proposed algorithm demonstrated high performance in terms of
accuracy, robustness and interpretability, producing results that are both reliable and actionable
for decision-makers.
The findings confirm that combining intelligent optimization techniques, particularly metaheuristic
algorithms, with explainable models provides a promising pathway for the
development of advanced diagnostic systems. This approach ultimately enhances the efficiency
and reliability of PV technologies while laying a solid foundation for real-world applications in
monitoring, fault detection and data-driven decision-making.