Diurnal Changes\nand Machine Learning Analysis of Perovskite\nModules Based on Two Years of Outdoor Monitoring

Long-term stability is the primary challenge for the\ncommercialization\nof perovskite photovoltaics, exacerbated by limited outdoor data and\nunclear correlations between indoor and outdoor tests. In this study,\nwe report on the outdoor stability testing of perovskite mini-modules\nconducted over a two-year period. We conducted a detailed analysis\nof the changes in performance across the day, quantifying both the\ndiurnal degradation and the overnight recovery. Additionally, we employed\nthe XGBoost regression model to forecast the power output. Our statistical\nanalysis of extensive aging data showed that all perovskite configurations\ntested exhibited diurnal degradation and recovery, maintaining a linear\nrelationship between these phases across all environmental conditions.\nOur predictive model, focusing on essential environmental parameters,\naccurately forecasted the power output of mini-modules with a 6.76%\nnRMSE, indicating its potential to predict the lifetime of perovskite-based\ndevices.