Methodology of Remote Diagnostics of Hot Jet Parameters Based on the Application of Deep Learning

This paper is devoted to the development and testing of a new approach to diagnostics of high-energy flows (in particular, plasma in reactors), based on the use of artificial neural networks. The main problem of traditional diagnostic methods is the impossibility of direct contact measurement of temperature profiles and concentrations of chemical species in high-temperature flow. In this regard, a method for the remote spectral measurement of flow thermal radiation is proposed.

This paper proposes an inverse radiation model based on an artificial neural network that is capable of extracting information about the temperature and concentrations of plasma components from infrared spectrum analysis. A radiation calculation technique is also presented, taking into account all the main factors affecting the processes of radiation transfer in plasma. Studies of the model have shown that the proposed approach demonstrates sufficient accuracy and potential for further development, although there is a need to refine the model for specific practical applications.