APPLICATION OF CLUSTER ANALYSIS METHODS FOR IDENTIFYING ACOUSTIC ACTIVE ZONES

Abstract

The article presents the results of a study on the application of cluster analysis methods to identify acoustically active zones (AAZ) in hazardous mining sites. Data obtained using the "Prognoz-ADS" seismic-acoustic system were utilized to analyze destruction processes in the rock mass, which manifest as a high concentration of acoustic emission events. The primary analytical tool used was the DBSCAN algorithm, which allows for the identification of clusters with arbitrary shapes, considering the spatial and temporal characteristics of the events. To optimize the algorithm's parameters, the Davies-Bouldin Index was applied, enabling the determination of the optimal neighborhood radius (7 m) and the minimum number of neighbors (13). As a result of analyzing data from the Yuzhnoe deposit, 20 acoustically active zones were identified, representing areas of elevated rock pressure. The analysis of clusters based on their spatial and temporal characteristics enabled the exclusion of events related to localization errors and anthropogenic noise, as well as the identification of patterns in the development of destruction foci. The developed software and methodological tools ensure accurate identification of destruction zones, precise delineation boundaries, and monitoring of their evolution. This study's practical value stems from its ability to forecast and prevent hazardous dynamic manifestations of rock pressure. The identified zones are recommended for priority monitoring to minimize the risks of emergency situations. The proposed approach has demonstrated its versatility and can be applied to analyze seismic-acoustic data from other hazardous mining sites. This work highlights the importance of integrating modern clustering methods with geomechanical monitoring tools to enhance the safety of mining operations.