Analysis of hybrid tandem solar cell using neural network

Abstract

In the 21st century, revolutions in the renewable energy sector have been accelerating at a rapid pace. It is due to the search for environment-friendly sources of energy and to fit the modus operandi of a sustainable society intended for the well-being of future generations. The solar energy sector is witnessing several attempts to increase the efficacy of solar energy systems. The ongoing research in Quantum Dot Solar Cells (QDSC), Perovskites and Carbon nanotubes leads to the advancement in natural energy. The QDSCs are a fairly interesting concept that aims to increase the efficiencies and costs of the solar cell. In this paper, the construction and working of a Tandem Solar Cell is analysed and the different design parameters are calculated. The cell is a monolithic hybrid integrated structure consisting of a low bandgap Colloidal Quantum Dot (CQD) back cell on top of an Organic Photovoltaic (OPV) cell. The hybrid tandem solar cell is simulated using MATLAB software and the various parameters are calculated using various experimentations. Further, a Levenberg-Marquardt training algorithm is used to train the experimented parameters for hybrid tandem solar cells. The training algorithm has been successfully used to accurately predict values for short circuit current density using three input parameters namely the open-circuit voltage and the thicknesses of the OPV and CQD, respectively. The various experimentation using ANN show that the experimental values are relatively matched with predicted results.