The methodology of evaluation and quality management of components of the urban ecosystem, based on the use
of new information methods, substantiation of marker biological objects and methods for assessing their response to
anthropogenic impact, and ways of data arrays folding used for evaluation has been developed. The presented methodology takes into account the processes of receipt, accumulation and response of biological objects to the effects of marker
substances. The approach for the simultaneous recording and analysis of heterogeneous and different-sized data having
different degrees of influence on the state of the studied urban ecosystem components using the cluster analysis method
is presented. The territory of the urban ecosystem was zoned using the neural self-organizing networks of Kohonen,
implemented in the author’s model of neural network filtering. A method for determining the probabilities of exceeding
the threshold values of concentrations of marker substances-metals in the studied components of the urban ecosystem
is proposed. We estimated the probability that in the territory to which a particular observation belongs, an excess of a
given threshold will be marked by at least one of the studied indicators. As thresholds for these events, we have taken
the medians of continuous series of observations of metal concentrations. As a result, we calculated the probabilities of
exceeding half of the studied parameters of the specified thresholds in each zone. The facts of exceeding the thresholds
(events) established in this way were recorded in the totality of observations selected in the form of clusters formed by
neural network classification. At the same time the spatial distribution of selected observations with specific address
bindings forms the corresponding territorial zones. Recommendations on the use of the obtained results for operational
and planned quality management of the urban ecosystem components at the local level are given.
Keywords: urban ecosystem, risk assessment, response of the human body, biosubstrates, neural network technologies, metals
Article published in number 4 for 2019 DOI: 10.25750/1995-4301-2019-4-031-036
ISSN 1995-4301
Select viewing options
