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

The construction of transparent mines is critical to the implementation of intelligent coal mining in China. Limited by current theories and technologies of geological exploration and multi-field fusion-based identification, the precision, 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 models. A dynamic identification technical framework for mine geological hazard risks is proposed, comprising five core components: 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 enables continuous perception and intelligent decision-making for mine geological hazard risks. In addition, the key techniques 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-related geological identification, water-rich geological identification, and dynamic identification of geological structures under disturbance conditions. The study provides new theoretical and technical support for building an intelligent, transparent, and reality-based transparent mine geological cloud platform.