The article is devoted to the development of an innovative neural network decision support system for firefighting in conditions of limited visibility. A comprehensive approach based on the integration of data from multispectral sensors (lidar, ultrasonic phased array, temperature and hygrometric sensors) is presented. The architecture includes a hybrid network that combines three-dimensional convolutional and bidirectional LSTM neurons. To improve the quality of processing, a cross-modal attention mechanism is used to evaluate the physical nature and reliability of incoming signals. A Bayesian approach is used to account for the uncertainty of forecasts using the Monte Carlo dropout method. Adaptive routing algorithms allow for quick response to changing situations. This solution significantly improves the efficiency of firefighting operations and reduces the risk to personnel.

Keywords: mathematical model, intelligence, organizational model, gas and smoke protection service, neural networks, limited visibility, fire department, management, intelligent systems, decision support

The article discusses the conceptual foundations of the transformation of the fire extinguishing management system based on the theory of complex organizational systems. The author substantiates the need to move from linear-hierarchical models to adaptive and networked structures capable of providing high stability and efficiency of response in conditions of uncertainty and dynamics of emergency situations. The analysis of the compliance of the fire extinguishing system with the characteristics of a complex organizational system has been carried out, contradictions between its complex nature and primitive control mechanisms have been identified, the causes and consequences of this paradox have been identified. Multi-agent digital platforms, the use of digital twins, situation centers, as well as the use of game theory methods to optimize resource allocation and decision support are proposed as ways to solve the identified problems.

Keywords: system approach, organizational system, firefighting, network structures, management, digitalization, transformation, game theory, optimization, criteria

The article presents the development of an analytical hydraulic model of the side branch pipe of a three-way fire branch DN80. The relevance of the work is due to the need for accurate hydraulic calculation of fire water supply systems, the effectiveness of which directly depends on the correct assessment of pressure loss in the fittings. The model is based on the method of element-by-element calculation, which takes into account local hydraulic resistances in areas with a sharp change in the flow geometry. The model includes losses at four bends (three at 45° and one at 90°), a straight-flow valve, and sudden expansion and contraction sections. The contribution of linear head losses, similar to the central branch pipe, was considered insignificant, accounting for only about 6% of the local losses. This confirms the key role of local resistances in determining the overall head loss in this element. The model was verified by comparing the calculated data with the results of experimental studies. It has been established that the discrepancy between the theoretically calculated value of the head loss (6.86 m) and the experimental value (6.97 m) is minimal, with a difference of only 1,6 %, indicating the high accuracy and adequacy of the developed model. The key practical result of the study is the calculation of the total coefficient of local resistances for the lateral branch pipe of the fire branch, which was found to be ζ = 3.4. This value can be directly used for simplified and accurate hydraulic calculations of pump-hose systems with three-way branches.

Keywords: Model, fire three-way junction, hydraulic resistance, head, pressure, flow rate