Original article Predicting the kinetics of complex luminescence processes in Python

Abstract

Introduction. Рolyarylene phthalides (PAF) are widely used in optoelectronics today. The reactions occurring during the synthesis of polyarylene phthalides have a complex character, which has not yet been described using mathematical models. In this regard, it is impossible to use PAF in many processes. Рolyarylene phthalides have luminescence, good optical and electrophysical properties. The elucidation of the mechanisms of the occurrence of luminescent states of PAF is of both fundamental and practical interest. The elucidation of the mechanisms of the occurrence of luminescent states of PAF is of both fundamental and practical interest. Due to the complexity of calculating the kinetics of the luminescence intensity of polyarylene phthalides using known mathematical models, the aim of the study was to build a system using machine learning methods that predicts luminescence values depending on temperature and heating time.Materials and methods. Experimental data have been prepared for calculations, the use of “random forest” and “gradient boosting” methods has been justified, a method for selecting hyperparameters of these models has been selected and the expediency of its use has been justified, optimal models have been constructed and predictions have been obtained.The results of the study. An algorithm for predicting the luminescence intensity of polyarylene phthalides has been developed. Using machine learning methods based on experimental data, the key hyperparameters of the system were determined and the average accuracy of predicting values was achieved — 80 %.Discussion and conclusions. High-accuracy forecasts will allow predicting how products containing polyarylene phthalides will react to external influences. The paper presents two methods for solving the problem, as they showed the best results.