Identifying rockburst-prone zones and their temporal evolution using energy-weighted MS visibility graphs

To improve the spatial characterisation of potentially hazardous zones from microseismic(MS) activity, this study proposes an energy-weighted MS visibility-graph method for hazardous-zone identification and graph-entropy-based precursor analysis. Log-transformed MS energy sequences are mapped onto visibility networks, and the mean energy-weighted clustering coefficient of local subnetworks is used to delineate potential hazardous zones through engineering back-analysis. Graph entropy within these spatially anchored zones is subsequently analysed to identify temporal precursors to high-energy events. Application to the 250101-2 longwall panel at Huating Coal Mine shows that high-value visibility-graph zones are spatially consistent with the locations of subsequent high-energy events, while graph entropy generally decreases before their occurrence. The hazardous zones exhibit a characteristic spatiotemporal evolution from dispersed to clustered distributions and from weak to strong anomalies. Validation under different mining conditions and at an additional longwall panel further demonstrates the applicability and stability of the method. These findings provide a simple, interpretable and readily implementable approach for identifying hazardous zones and issuing short-term warnings of high-energy MS events in coal mines. 1. An energy-weighted microseismic visibility graph method is developed for spatial identification of rockburst-prone zones. 2. VG high-value zones and normalized graph entropy are combined to link hazardous-zone location with short-term precursor evolution. 3. A spatiotemporal precursor pattern of hazardous-zone evolution is revealed and validated under different mining conditions and longwall panels.