Prediction of device performance in SnO2 based inverted organic solar cells using Machine learning framework

Abstract

The development of wearable electronic gadgets has spanned the research attention toward the design of flexible and high-performance organic solar cells. The complicated process and long data execution time have limited its research progress. In this project, the machine learning (ML) framework with different algorithm models and kernel functions was employed to predict the device performance of solution-processed SnO₂-based organic solar cells. The device performance of the SnO₂ prepared using different spinning rates was used as the training data for machine learning prediction. The accuracy of the prediction was controlled using the root-mean-square error (RMSE), mean absolute percentage error (MAPE), and coefficient of determination (R²). The comparison between the measured and predicted value of the device parameters such as open circuit voltage (V_oc), short circuit current density (J_sc), fill factor (FF), and power conversion efficiency (PCE) was discussed. The radial basis support vector regression (SVR) integrated with particle swarm optimization (PSO) model showed the highest performance in predicting the PCE of SnO₂-based organic solar cells with R² of 99%, RMSE of 0.0119 and MAPE of 0.0075. This novel study demonstrated that support vector regression (SVR) integrated with the particle swarm optimization (PSO) model is an alternative method to predict the device performance in future organic solar cells.