Artificial neural networks for predicting optical conversion efficiency in luminescent solar concentrators

•Spectral converting devices enhance solar energy conversion efficiency significantly.•Machine learning empowers material design without physical implementation.•ANN model predicting optical conversion efficiency without physical experiments.•Model predicts performance with uncertainties smaller than experiments.•ANN-based decisions foster a sustainable energy future. Developing light-harvesting materials able to shape the sunlight to cope with the absorption region of photovoltaic (PV) cells presents an opportunity for the utilization of spectral converters like the luminescent solar concentrators (LSCs). This study explores the use of artificial neural networks (ANNs) to predict the optical conversion efficiency of spectral converters, based on the material properties employed in their production, without the need for expensive and time-consuming experimental testing. To predict efficiency as a function of materials and manufacturing processes, ANNs were trained using data from previously documented physical implementations. The findings indicate that ANNs, having 97 and 19 neurons in the hidden layers, provide accurate efficiency predictions, making them a valuable tool for designing and optimizing spectral converting systems. The proposed model was validated and got a mean square error in the order of 10−5 for the optical conversion efficiency. The trained ANN introduced a novel methodology for predicting the optical efficiency of spectral converters, opening the door to the application of machine learning as a decision-making tool for material design, and eliminating the necessity for physical device implementations.