3D mapping of particulate matter in the occupational environment via low-cost sensor network and machine learning

Effective management of occupational exposure to airborne particles in industrial environments requires spatially resolved monitoring tools capable of capturing the transient dynamics of dust dispersion. Conventional point measurements and gravimetric methods, while regulatory-compliant, fail to provide the three-dimensional spatial information needed to identify high-exposure zones and design targeted prevention strategies. This study presents a Machine Learning enhanced model for three-dimensional reconstruction of particle concentration fields in occupational environments, based on a network of ten calibrated low-cost SPS30 sensors coupled with an XGBoost regression algorithm and a hybrid triangulation/Inverse Distance Weighting interpolation scheme. The model was assessed through a progressive experimental strategy of increasing complexity: simulation, controlled laboratory aerosol generation, pilot stone cutting bench, and real industrial conditions. In the laboratory and pilot configurations, the model achieved a coefficient of determination R 2 consistently above 0.75 (reaching 0.91–0.98 under controlled conditions) and a Mean Absolute Percentage Error below 20% for number concentrations below 3000 particles/cm 3 , demonstrating satisfactory predictive capacity even under pulsed emissions and complex airflows. The industrial-scale trial is reported as a demonstration of applicability and reproduced the asymmetric plume propagation observed during manual finishing operations. The resulting interactive 3D concentration maps, displayed with user-defined exposure thresholds, enable the identification of high-exposure zones and the characterization of dust plume evolution over time. Sensors' limitations were identified above 3000 particles/cm 3 , beyond which measurement uncertainty increases significantly. This approach constitutes a promising, flexible, and cost-effective tool for proactive respiratory risk management in the natural stone sector and transferable to other dust-generating industries.