Neural Network as a Cost Function for EPSO Algorithm in Perovskite Solar
Cell Simulation
Abstract
Time-consuming is one of the main bottlenecks in the SCAPS-1D
simulations in the case of more computational data set. In this regard,
we are achieving outputs data in SCAPS-1D and repeat this simulation by
employing a neural network as a cost or target function for the
evolutionary particle swarm optimization (EPSO) algorithm to decrease
the computational expensiveness of SCAPS-1D simulation. Optimization and
numerical simulation tools pave the way for having a better insight into
the designing of perovskite solar cells. Also, it allows finding a
relation between artificial intelligence and device physics.