Gases and mixtures explosiveness estimation using a model trained by limited sets of gases

Prevention of emergencies associated with flammable gases explosions remains an urgent task. A ranking place in solving of this task plays environment monitoring for combustible gases presence. The environmental explosiveness level can also be assessed overall, without measuring separate gases concentrations. The assessment of environmental explosiveness level can be performed using catalytic gas sensors based on combustion heat measurements of flammable gases combustion in the sensors. Modern data processing methods application, such as machine learning, allows to increase the measurements accuracy. However, machine learning needs a Iot of data. To collect that data enough measurements for different gases should be performed. At the same time, the question remains of whether it is possible to apply the machine learning models to gases, which have not been used while training. In this work, the results of the research are presented, where the possibility of successful models training using limited gases set and the ability of such models to assess the explosiveness level of other gases and mixtures were examined.