Condition Monitoring of Photovoltaic Panels Through Electrical Impedance Spectroscopy and Machine Learning Focusing on Temperature, Dust, and Microcracks

The efficiency of Photovoltaic (PV) systems depends on environmental conditions that include temperature variations, dust accumulation and microcracks. These factors severely affect the output power of PV systems particularly where PV is installed domestically for which monitoring and maintenance are challenging due to cost constraints. Traditional numerical schemes are limited because PV systems exhibit nonlinear behavior due to environmental factors. To address this issue, electrical impedance spectroscopy (EIS) combined with machine learning is studied, modeling different scenarios to monitor the impact of temperature variations, dust and microcracks on photovoltaic (PV) systems.The experiment was conducted in a controlled environment, generating datasets for machine learning using EIS on solar panels through direct current (DC) lines over a frequency range of 1 Hz to 25 MHz. The light gradient boosting machine (LGBM) demonstrated great performance, achieving an F1 score of 0.856 for temperature classification and 0.975 for detection of dust and microcracks. The results of this study suggest the integration of EIS with machine learning schemes presents a cost effective and viable solution for real time monitoring of PV systems. This approach not only mitigates the effect of environmental factors but also reduces the cost of maintenance and increases predictive maintenance ensuring enhanced PV system performance.