A group improvised PSO-random forest-based intelligent hybrid approach for advancing perovskite solar cell efficiency

Abstract

Perovskite solar cells (PSCs) have gained significant attention in the solar industry owing to their excellent optical absorption and low charge recombination rates, thereby promising high power conversion efficiency (PCE). Traditional approaches in material studies, such as characterizations and ground state energy determination, using density functional theory (DFT) calculations and prefabrication simulations have limitations in terms of time, cost, and computational complexity. An intelligent hybrid approach was therefore proposed for fabricating highly efficient PSCs in relatively less time, with less computational complexity and better cost effectiveness. The methodology included a hybrid approach of random forest regressor (RFR) in conjunction with group improvised particle swarm optimization (GIPSO). This strategy offered a better way for studying the performance of PSCs with different layers and contact parameterizations to achieve an optimized design. The adaptive design of perovskite solar cells using the intelligent hybrid optimization and prediction of PSC structure depended on the factors of efficiency, fill factor, short-circuit current density and open-circuit voltage. The methodology intelligently hybridized RFR with GIPSO techniques to achieve highly efficient PSCs, which were further examined through SCAPS-1D simulations. Experimental results demonstrated that the optimized PSC structure achieved high efficacy comparable to those of state-of-the-art cells, supporting the potential of the proposed approach to develop better-performing perovskite cells.