Detection and classification of failure types of solar panels using machine learning with I-V curve parameters
Conference proceedings article
Authors/Editors
Strategic Research Themes
Publication Details
Author list: Kittipob Wiriyavorawet, Manit Seapan, Dhirayut Chenvidhya, Panusorn Polchai, Tanokkorn Chenvidhya, Yaowanee Sangpongsanont, Chamnan Limsakul
Publication year: 2026
Title of series: PVSEC-36, 2025 PROCEEDINGS The 36th International Photovoltaic Science and Engineering Conference
Number in series: -
Volume number: -
Start page: 101
End page: 102
Number of pages: 2
Abstract
A large number of long-term solar panels are now in operation, analysing module degradation is essential for predicting potential failures and planning effective maintenance. This study evaluates electrical performance parameters measured under Standard Test Conditions (STC) and compares them with datasheet specifications including short-circuit current (Isc), open circuit voltage (Voc), maximum power (Pmax), current at maximum power (Imp), voltage at maximum power (Vmp), shunt resistance (Rsh), and series resistance (Rs) to identify characteristic failure patterns. By comparing current–voltage (I–V) curves from panels that have been in field operation for extended periods, meaningful insights can be obtained even when pre-installation I–V data are unavailable. Raw I–V data from CSV files are processed using Python-based machine learning to classify failures into three categories: Rsh-slope degradation, Isc degradation, and combined multi-mode failures. Parameter deviations from the datasheet are used as model inputs to accurately distinguish among these failure types.
Keywords
Failure mechanism, Machine Learning, Solar photovoltaic






