Mining operations release large quantities of particulate matter as coal and rock strata are disrupted by drilling, blasting, and mechanical excavation. The physical (e.g., size and morphology) and chemical (e.g., ash and sulfur content) properties of these particles are highly correlated with the in-situ characteristics of the coal-rock body. Real-time acquisition of spectral, image, and compositional data from these particles enables rapid identification of geological materials, providing dynamic feedback for intelligent mining. However, particle dispersion and the presence of flammable gases such as methane in fully mechanized mining faces pose safety and technical challenges for conventional analysis methods. Here we report a compact, explosion-proof sampling and analysis device based on aerodynamic separation. The system integrates a negative-pressure fan, cyclone separator,? pressure sensor, and control module for sampling, transport, and analysis. We use a computational fluid dynamics–discrete phase model (CFD-DPM) to simulate and optimize the internal particle separation process, with separation efficiency and total pressure drop as the primary metrics. The device achieves over 95% separation efficiency under sealed negative pressure. CFD-guided structural design improves classifier stability and transport continuity. A two-axis platform transfers quantified particles into a sealed chamber for rapid spectral and compositional analysis. This system enables adaptive, in-situ coal-rock discrimination and offers a scalable solution for intelligent, unmanned mining operations.