Determining temperature and partial pressures of the components of a high-temperature gas mixture using artificial neural networks

Abstract

The article deals with solving the inverse problem of gaseous media optics by determining the parameters of high-temperature gaseous media from its transmittances using the artificial neural networks. The study of the dependence of the maximum relative error in determining the desired parameters on the size of the training set and the artificial neural network configuration is carried out. The possibility of solving the inverse problem in the case of a four-component gas mixture (water vapor, carbon dioxide, carbon oxide and nitrogen oxide) is shown.