BUILDINGS FIRES FREQUENCY DETERMINING METHODOLOGY BASED ON DENSITY ESTIMATION AND SIMULATED ANNEALING METHODS

Abstract

Solving the problem of determining the optimal spatial location of fire departments is a rather complex
scientific and technical problem, including, as previous studies have shown, an extensive list of factors,
including the need to assess the expected frequencies of fires in different parts of settlements, depending
on the nature of the development. Currently, in the Russia, approaches and methods to solve this problem
are not sufficiently developed. As a rule, researchers limit themselves to the fact of the existence of a
spatial distribution of fires, without delving into the causes that led to one or another nature of such a
distribution. Meanwhile, their understanding will allow us to build models for estimating the expected
densities of fire flows in various areas. The article proposes an approach based on the method of estimating
the spatial density of random events (KDE, Kernel Density Estimation) and a simulated annealing
algorithm to select the values of the calculated frequencies of fires in buildings of various classes of functional
fire hazard. The approach has been tested on the available data on fires for the period 2010-2020
and urban development in the city of Krasnoyarsk. The study showed that the proposed approach allows
us to obtain such values of fire occurrence frequencies at which their predicted density will be as close as
possible to the actual one. The results obtained expand the set of research tools in the field of assessing
both the actual and predicted fire situation and are aimed at developing methods and algorithms for determining
the optimal locations of fire departments. The proposed approach can also be used to solve
other problems of spatial optimization in the field of public safety, road safety, protection of the population
from emergency situations, as well as in the field of urbanism and urban planning