摘要：Harvesting the energy from the sun through the bulk heterojunction (BHJ) solar cells need materials with specific electronic characteristics. However, any promising material if cast improperly in cells will end into low or even null power conversion efficiency (PCE). Cell casting optimization is a time/material consumable step in any photovoltaic manufacturing practice. In this study, we showed that how the artificial intelligence (AI) could help to find optimum values of device preparation variables. For this purpose, an in-house code will catch the input variables (donor: acceptor ratio, spin casting rate, annealing temperature); learn the trends by the hybrid artificial neural network (ANN), genetic algorithm (GA) and optimize the output results simultaneously. The results showed that ANN/GA is capable to learn the trends of relatively small size dataset without over-fitting. This study highlights that how implementing the suggested AI model can help to learn more information and find the optimum recipe from less number of experiments with the highest precision.
关键词：bulk heterojunction solar cells;Artificial Neural Network;Genetic Algorithm;Multi-Objective Optimization
