Characteristic gas distribution-based algorithm for conveyor belt fire source localization in mine tunnels

Fires originating from belt conveyors present significant hazards in coal mines, necessitating prompt and accurate localization of fire sources. This study proposes a novel fire source location algorithm predicated on the distribution patterns of characteristic gases within coal mine tunnels. This method combines a three-dimensional Gaussian plume gas diffusion model applicable to confined tunnel spaces with a hybrid GA-PSO (genetic algorithm particle swarm optimization) algorithm. To validate the approach, the conveyor belt combustion experiments on the small-scale tunnel were performed, complemented by numerical simulations, to characterize the spatiotemporal distribution of HCl (Hydrogen Chloride) during combustion. In practical application, HCl concentration data from sensors are utilized in an inversion calculation, leveraging the optimized diffusion model to determine the precise fire source coordinates. Findings indicate that the relative error of the predicted location information is small, the error is reduced by 87.5–91.7 % compared with other algorithms, and the location accuracy is more than 98 %, validating its effectiveness and feasibility for fire source identification in the challenging subterranean conditions of coal mines. This work strengthens the mechanistic link between characteristic-gas evolution and fire location, providing a quantitative tool to optimize sensor layout, accelerate fire detection, and support more timely evacuation and emergency response in underground coal mines.