Method of reducing fire hazard of technological process in the event of oil spillage based on mathematical models of neural network.

The study provides a theoretical justification of method applied for reducing the fire hazard at oil and gas enterprises through monitoring the production area with unmanned aerial vehicles equipped with automated identification systems. The identification problem can be solved using mathematical models of neural networks. The paper presents the existing and promising practices of visual observation through various flying platforms. Three types of events under analysis have been selected that affect fire safety: oil and petroleum products spill, flame combustion, and smoke. The framework conditions for unmanned aircraft flight are described. Reconciliation and recurrent mathematical models of neural activity were selected as the most suitable mathematical models of image processing. An example of image mapping received through a drone is provided where three types of identifiable events are present. Conclusions are made about the applicability of mathematical models of neural networks for identifying fire hazards. [ABSTRACT FROM AUTHOR]