ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING

The study of the perovskite solar cell parameter influence on their efficiency by using machine learning

I. E. Novoselov^a, A. A. Smirnov^a, I. S. Zhidkov<sup>b

a</sup> Ural Federal University named after the First President of Russia B. N. Yeltsin, Ekaterinburg
^b Institute of Metal Physics, Ural Branch of the Russian Academy of Sciences, Ekaterinburg

Abstract: This paper presents an approach to optimize perovskite solar cells (PSC) by using machine learning algorithms. The aim of the study is to find the best PSC parameters that provide high power conversion efficiency (PCE). To train the algorithms, the dataset with PSC feature information (the perovskite layer thickness, the electron transport layer type, the target variables, etc.) was created. Various machine learning algorithms were applied, the XGBoosting, CatBoost and Random Forest showed the low error in both learning stages (regression where the target variable is PCE, multi-target regression where the four variables are PCE, Voc, Jsc and FF as well). The most important variables for PCE are Voc, Jsc and FF, the less important ones – Pero th, ETL, Cs, MA, FA, I and HTL.

Keywords: perovskite solar cells, machine learning, optimization, power conversion efficiency, PCE.

DOI: 10.60797/IRJ.2024.143.121