The releases of toxic or explosive gases have a negative impact on the environment and pose a serious threat to the life and health of industrial workers, as well as to the population living in areas adjacent to the industrial facilities. Modern technologies make it possible to remotely and promptly detect such threats, thereby preventing potential accidents and disasters. This work presents a novel methodology for simulating the detection of a gas cloud resulting from a leak at an industrial infrastructure line under open atmospheric conditions. The approach includes the synthesis of observation scenarios in the radiation wavelength range of 300–2500 nm, taking into account the peculiarities of its detection utilizing hyperspectral imaging instrumentation (HSI). Using the example of sulfur dioxide leak detection via a neural network algorithm based on a Siamese neural network, it has been demonstrated that an SO cloud can be remotely identified using HSI operating in the 330–700 nm range with a spectral resolution of 1 nm.
I. D. Rodionov (Wed,) studied this question.