Calculation of time of evacuation of people using the forecasting method based on use of molecular descriptors and artificial neural networks

Abstract

For calculation of time of evacuation of people, according to the technique stated in GOST 12.1.004-91 such fire-dangerous indicators of substances are necessary: specific mass speed of burning out, linear speed of distribution of a flame, the lowest warmth of combustion. In most cases basic data in reference and standard books are absent. In such cases it is recommended to use the substance parameters similar on a structure to initial substance. In such results there is a big error which doesn't provide an integrated approach to development of the system of prevention of the fire. For optimization of a choice of the actions aimed at providing fire safety it is offered to use the forecasting method based on use of molecular descriptors and artificial neural networks. By means of a method, the absent parameters for a dipropilketon were predicted. As a result of approbation of this approach by calculation of time of evacuation of people on the example of limit ketones of the dipropilketon and benzene which is a part of the first calculation was carried out. Comparing the received results, it is visible that impact of dangerous factors of the fire on the person come in case of use of the predicted values that provides more rigid approach in ensuring fire safety. GOST 12.1.004-91 technique in a complex with the forecasting method based on use of molecular descriptors and artificial neural networks I proved from a positive side and allows to time evacuations of people with the acceptable accuracy.