| Abstract | Ensuring functioning of the enterprises demands knowledge and performance of existing rules and rules of labor protection and fire safety, and possibility of forecasting and minimization of risks of emergence of technogenic emergencies. Such providing assumes development of the express methods of the analysis allowing predicting physical and chemical properties of substance. Fire and explosive safety properties have special value. The techniques existing now determined by GOST 12.1.044-89 became outdated and demand concentration of attention when carrying out difficult experiments by detection of fire safety substances properties. For the solution of this task, two methods of forecasting offered: settlement and based on sharing of descriptors and artificial neural networks. In research, approbation of these methods on the example of forecasting of the minimum energy of ignition of limit aldehydes carried out. Thus, the error of a calculation method made 0,07 mJ, the method based on sharing of descriptors and artificial neural networks 0,01 mJ. Perspective use of the method based on use of descriptors and artificial neural networks shown.
|
