Enhancing Photovoltaic System Resilience: A Logistic Regression Approach to Fault Diagnosis

Addressing the imperative need for fault diagnosis in Photovoltaic (PV) systems, this study presents a fault detection method utilizing the Logistic Regression classifier. PV systems, which are integral to sustainable electricity generation, are susceptible to faults over time, resulting in efficien...

Full description

Saved in:
Bibliographic Details
Published in2024 International Conference on Recent Advances in Electrical, Electronics, Ubiquitous Communication, and Computational Intelligence (RAEEUCCI) pp. 1 - 5
Main Authors Ragul, S., Tamilselvi, S., Elamurugan, P., Bharathidasan, S.
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.04.2024
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:Addressing the imperative need for fault diagnosis in Photovoltaic (PV) systems, this study presents a fault detection method utilizing the Logistic Regression classifier. PV systems, which are integral to sustainable electricity generation, are susceptible to faults over time, resulting in efficiency degradation. To counter this challenge, an optimized machine learning algorithm is proposed for identifying open circuit and short circuit faults. Here, the dataset, obtained through MATLAB simulations under diverse fault conditions, is pre-tuned, obviating the requirement for post-simulation hyper-parameter tuning. Partitioned into an 80% training subset and a 20% testing subset, the dataset forms the foundation for training and validating the Logistic Regression classifier. Notably, the classifier exhibits promising accuracy in fault identification, positioning it as a viable tool for real-world fault detection scenarios. The study underscores the significance of fault diagnosis in ensuring the reliable and sustainable operation of PV systems, contributing to the long-term efficiency and effectiveness of these systems in electricity generation. For solar cells, we provide a PV EL Anomaly Detection (PVEL-AD) dataset of 36,543 near-infrared images with a variety of internal faults and a heterogeneous backdrop. The Logistic Regression classifier (shows 88.7 % accuracy) emerges as a robust solution for fault identification, offering practical implications for the enhancement of PV system performance.
DOI:10.1109/RAEEUCCI61380.2024.10547898