04268ntm a2200337 i 4500 MY-KuUP 20251230153400.0 t||||fr|||| 000 0 ta 251230t20252025my a|||fr|||| 000 0 eng d THE0010192 (Local) Hardback UMPSA eng UMPSA rda FTKEE .I43 2025 r Thesis Md Imamul Islam, author. Performance and aging phenomena assessment of field-aged solar pv modules in tropical climate: an integrated analysis using yolov9 and regression algorithms / Md Imamul Islam Kuantan, Pahang : UMPSA, 2025 © 2025 xvi, 173 pages : illustrationS ; 30 cm. + 1 CD-ROM. rdacontent text rdamedia unmediated rdacarrier volume rda text file PDF Faculty of Electrical & Electronics Engineering Technology Thesis (Master of Science) -- Universiti Malaysia Pahang – 2025 Include bibliographical reference Solar Photovoltaic (PV) systems in tropical climates face unique challenges stemming from diverse environmental conditions, requiring advanced methods for assessing performance and identifying aging phenomena. Traditional evaluation techniques, including visual inspection, I-V curve analysis, and thermal imaging, although useful, frequently necessitate specialized tools and can be time-intensive, which restricts their accuracy for extensive assessments. This thesis introduces a novel hybrid approach that combines You Only Look Once version 9 (YOLOv9)-based hotspot identification with I-V curve-based advanced machine learning technique to explore and predict aging phenomena in field-aged solar PV modules, effectively tackling the present limitations. While the methodology incorporates data collection through visual inspection, I-V curve analysis, and thermal imaging, it enhances these techniques by using the collected data to train machine learning models that can subsequently perform automated assessments with minimal equipment and expertise requirements. To achieve this comprehensive approach, a detailed research design was implemented, which included the collection of data from field-aged PV modules and environmental parameters. The electrical dataset was collected utilizing the I-V curve analyzer, which measured essential performance parameters such as open circuit voltage, short circuit current, maximum power current, and maximum power. The thermal dataset comprised more than 2,100 infrared images, which ensured a high level of accuracy in identifying thermal anomalies. The visual inspection yielded additional information regarding physical degradation and contamination. The gathered data underwent processing and analysis, resulting in the design of YOLOv9 for hotspot detection and the execution of a regression model. The findings of the field studies revealed significant degradation trends, showing power degradation rates of 6.48% for monocrystalline modules and 12.76% for polycrystalline modules, along with annual degradation rates of 0.81% and 1.27%, respectively. Visual evaluations indicated widespread concerns such as the yellowing of Ethylene Vinyl Acetate (EVA) affecting nearly 90% of panels, along with delamination and accumulation of dust. The YOLOv9-based deep learning model achieved a mean Average Precision (mAP50) of 0.93681, exceeding existing methods in terms of accuracy and efficiency. Furthermore, the machine learning-based ridge regression model demonstrated impressive accuracy, attaining a Mean Squared Error (MSE) of around 0.99% across various regression models. This thorough approach offers a more efficient and accurate assessment of PV module depreciation in tropical areas. . The findings significantly improve the understanding of PV module degradation in tropical climates and offer practical tools for enhancing the sustainability and reliability of solar energy systems in challenging weather conditions. Faculty of Electrical & Electronics Engineering Technology Dissertations Universities and colleges Dissertations Theses Dissertations lcc THESIS 0 0 lcc 0 1 20000 20000 2025-12-30 0 FTKEE .I43 2025 r Thesis T000003895 2025-12-30 15:34:14 2025-12-30 THESIS 0 0 lcc 0 -2 20000 20000 2025-12-30 0 CD13818 T000003896 2025-12-30 15:35:01 2025-12-30 THESIS 103811 103817