Photovoltaic conversion of organic photovoltaic materials based on molecular fingerprint feature engineering

In recent years, organic solar cells (OSCs) have gradually become the focus of renewable energy research. In order to predict the photovoltaic characteristics of OSCs more accurately and efficiently, researchers have incorporated numerous machine learning models into their studies. In this research, we designed and implemented a neural network model based on molecular fingerprints and applied it to the study of the power conversion efficiency (PCE) prediction of OSCs. We explored the impact of different organic photovoltaic material structures and feature extraction methods on the prediction of the PCE. Through experimental evaluations, the model not only achieved good experimental results in terms of material structure and PCE prediction but also compared the feature extraction methods of different molecular fingerprints. It was found that both Morgan and Circular fingerprints performed excellently in multiple scenarios.