Electrical Faults Analysis and Detection in Photovoltaic Arrays Based on Machine Learning Classifiers

Fouad Suliman (), Fatih Anayi and Michael Packianather
Additional contact information
Fouad Suliman: School of Engineering, Cardiff University, Cardiff CF24 3AA, UK
Fatih Anayi: School of Engineering, Cardiff University, Cardiff CF24 3AA, UK
Michael Packianather: School of Engineering, Cardiff University, Cardiff CF24 3AA, UK

Sustainability, 2024, vol. 16, issue 3, 1-29

Abstract: Solar photovoltaic energy generation has garnered substantial interest owing to its inherent advantages, such as zero pollution, flexibility, sustainability, and high reliability. Ensuring the efficient functioning of PV power facilities hinges on precise fault detection. This not only bolsters their reliability and safety but also optimizes profits and avoids costly maintenance. However, the detection and classification of faults on the Direct Current (DC) side of the PV system using common protection devices present significant challenges. This research delves into the exploration and analysis of complex faults within photovoltaic (PV) arrays, particularly those exhibiting similar I-V curves, a significant challenge in PV fault diagnosis not adequately addressed in previous research. This paper explores the design and implementation of Support Vector Machines (SVMs) and Extreme Gradient Boosting (XGBoost), focusing on their capacity to effectively discern various fault states in small PV arrays. The research broadens its focus to incorporate the use of optimization algorithms, specifically the Bees Algorithm (BA) and Particle Swarm Optimization (PSO), with the goal of improving the performance of basic SVM and XGBoost classifiers. The optimization process involves refining the hyperparameters of the Machine Learning models to achieve superior accuracy in fault classification. The findings put forth a persuasive case for the Bees Algorithm’s resilience and efficiency. When employed to optimize SVM and XGBoost classifiers for the detection of complex faults in PV arrays, the Bees Algorithm showcased remarkable accuracy. In contrast, classifiers fine-tuned with the PSO algorithm exhibited comparatively lower performances. The findings underscore the Bees Algorithm’s potential to enhance the accuracy of classifiers in the context of fault detection in photovoltaic systems.

Keywords: photovoltaic systems; PV string; I-V curve analysis; Support Vector Machine (SVM); Extreme Grading Boosting (XGBoost); Bees Algorithm (BA); Particle Swarm Optimization (PSO); fault classification (search for similar items in EconPapers)
JEL-codes: O13 Q Q0 Q2 Q3 Q5 Q56 (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2071-1050/16/3/1102/pdf (application/pdf)
https://www.mdpi.com/2071-1050/16/3/1102/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jsusta:v:16:y:2024:i:3:p:1102-:d:1327965

Access Statistics for this article

Sustainability is currently edited by Ms. Alexandra Wu

More articles in Sustainability from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().