Year: 2024
Type: article
Abstract: The greatest influence on the destructive potential of natural disasters is the weather, which is changing due to global warming. Therefore, the presentation of operational forecasts of various lead times, indicating the probable risks of incidents (emergency situations) is one of the main tasks of modern crisis management centers of territorial bodies of the constituent entities of the Russian Federation. In this regard, the authors of the work determined the goal of the study – the development of a predictive risk management model using the most powerful technological tool of artificial neural networks. To achieve this goal, a mathematical algorithm for the operation of neural networks with backpropagation of errors was proposed, implemented by a special software product. Using a prepared training sample of meteorological observations, we simulated the work of artificial neural networks to predict dangerous weather phenomena and the likelihood of heavy rains. Analysis of the results obtained made it possible to establish the permissible value of relative and absolute error.
Authors: Denis Korolev, Elena Russkih, Sergey Konchakov
Language: English
Subfield: Environmental Engineering
Field: Environmental Science
Domain: Physical Sciences
SDG: Climate action
Open Access status: bronze
OpenAlex ID: W4405375068