354 More generally, the spatial risk indicator can be monitored over subsequent operational periods to evaluate the effectiveness of such preventive measures, supporting continuous, data-driven safety management aligned with operational risk interpretation. 5. CONCLUSIONS This work developed and implemented an adaptive multi-expert system, based exclusively on Collision Avoidance System (CAS) data, for contextual interpretation and prioritization of interaction alerts and spatial analysis of collision risk in mining operations. The approach integrates intelligent mine segmentation, a fuzzy expert system for severity estimation, and a spatial aggregation model for hotspot identification, forming a coherent architecture oriented toward a mine safety digital twin. The automatic segmentation identified functional zones with spatial accuracy above 90%, validated with safety and operations specialists, providing the contextual foundation for differentiated risk evaluation. The methodology was implemented within a Hexagon Mining CAS environment and evaluated using operational data from large-scale open-pit iron and copper mining operations in Brazil and Peru, demonstrating applicability in complex mining contexts. In the alert prioritization stage, the fuzzy expert system processed several thousand equipment interactions, including haul truck–light vehicle interactions, identifying a subset of high-severity interactions according to the contextual risk interpretation defined by the expert system. This demonstrates the framework’s ability to support operational teams in interpreting CAS alerts through contextual severity indicators aligned with site-specific safety criteria. Validation with a CAS specialist yielded sensitivity above 95% and precision around 90%, confirming strong agreement with expert interpretation of operational risk. The false negative rate remained below 5%, indicating robust performance under real operational conditions. The spatial aggregation analysis identified accumulated-risk zones and recurrent operational hotspots, particularly at mixed-traffic intersections involving haul trucks and light vehicles. These results reflect interaction patterns consistently observed across the analyzed operations and supported practical safety recommendations related to traffic control and intersection management, demonstrating the value of the approach for early detection of systemic risks and data-driven preventive actions. Overall, the framework provides an additional analytical layer that translates CAS interaction alerts into contextual safety indicators aligned with the operational risk perception of each mining site. In its current form, the system operates as a data-driven safety monitoring layer (digital shadow), providing continuous historical and near-real-time visibility of operational risk patterns and establishing the foundation for a predictive mine safety digital twin. REFERENCES EMESRT. (2023). Vehicle Interaction Control Improvement Project Guide (VICI Project Guide). Earth Moving Equipment Safety Round Table. EMESRT. (2024). Vehicle Interaction Leading Sites Program: Principles & Success Factors. Earth Moving Equipment Safety Round Table. Hexagon Mining. (2020, April 22). Inside Hexagon’s Collision Avoidance System. Hexagon Mining Blog. Hrica, J. K., Bellanca, J. L., Benbourenane, I., Carr, J. L., Homer, J., & Stabryla, K. M. (2022). A Rapid Review of Collision Avoidance and Warning Technologies for Mining Haul Trucks. Mining, Metallurgy & Exploration, 39(4), 1357–1389. https://doi.org/10.1007/s42461-022-00633-w Ruff, T. M. (2007). Recommendations for evaluating and implementing proximity warning systems on surface mining equipment (NIOSH Publication No. 2007–146; RI 9672). National Institute for Occupational Safety and Health. https://doi.org/10.26616/NIOSHPUB2007146
RkJQdWJsaXNoZXIy MTM0Mzk2