Dynamic fusion-based identification framework and key technologies for mine geological hazard risks

PENG Suping; XU Xianlei; MA Zheng; XI Yuhe; GENG Henggao

doi:10.13225/j.cnki.jccs.2025.1528

Abstract：The construction of transparent mines is critical to the implementation of intelligent coal mining in China.Lim-ited by current theories and technologies of geological exploration and multi-field fusion-based identification,the preci-sion,accuracy,and timeliness of existing mine geological hazard risk identification still cannot meet the requirements of transparent mine construction in the new era for the application of multi-source geological information data.On the basis of summarizing the development of techniques such as seismic-electrical joint inversion,borehole-surface joint inversion,multi-wave and multi-attribute joint dynamic inversion,and multi-source,multi-scale,multi-physical-field joint inversion,this paper proposes the concept of a"three-scale-three-resolution-one-model"transparent mine.The three scales refer to the mine area,the mining district(resolution of 3 m),and the working face(resolutions of 0.5 m and 0.1 m),while the"one model"refers to a transparent geological model that is unified in space and time.A multi-source heterogeneous data management scheme is designed,including standardized acquisition,cleaning,annotation,and lakehouse-based integrated management of multi-source and multi-modal mine data,providing highly reliable data support for geological large mod-els.A dynamic identification technical framework for mine geological hazard risks is proposed,comprising five core com-ponents:real-time perception and aggregation of multi-source and multi-field information,comprehensive integration and processing of multi-field geophysical data,multi-field joint inversion,feature construction and dynamic identification,and early warning,model updating and interaction with large models.Through a closed-loop workflow of"data sensing-data processing-multi-field inversion-feature construction-dynamic identification-feedback optimization",the framework en-ables continuous perception and intelligent decision-making for mine geological hazard risks.In addition,the key tech-niques for constructing typical mine geological hazard risk identification models are systematically elaborated,including six types of models:surface geophysical multi-field joint inversion and identification,in-situ while-drilling detection and identification in development roadways,coal-rock layering and structural identification at fully mechanized faces,gas-re-lated geological identification,water-rich geological identification,and dynamic identification of geological structures un-der disturbance conditions.The study provides new theoretical and technical support for building an intelligent,transpar-ent,and reality-based transparent mine geological cloud platform.