基于STSNN聚类算法的用沙坝矿微震事件活动特征研究

利用时空共享近邻聚类算法（STSNN）对用沙坝矿微震事件进行聚类分析，通过对噪声率进行有效性评价，最终确定k=6和ΔT=6为该算法的最佳输入参数，识别得到98个微震事件聚集区域，最大类簇有544个微震事件，并且该类簇主要集中在用沙坝矿的断层区域。对该类簇微震事件活动特征进行分析，主要包括微震事件的24 h分布、微震事件活动率、视体积、施密特数及劲度系数，根据活动规律的变化特征，提出微震活动率急剧下降并且累积视体积曲线忽然上升、施密特数和劲度系数先升后降的点作为岩体失稳发生破坏性事件的预警点。通过对微震事件活动规律的研究可为大事件的产生提供有效的预判信息，为保证矿山安全生产发挥重要的作用。

Study on activity characteristics of micro-seismic events in Yongshaba mine based on STSNN clustering algorithm

The cluster analysis was conducted on the micro-seismic events in Yongshaba mine by using spatio temporal shared nearest neighbor (STSNN) clustering algorithm. k=6 and ΔT=6 were determined to be the best input parameters of the algorithm through the noise rate and effectiveness evaluation. 98 micro-seismic events gathering areas were identified, and the biggest class cluster consisted of 544 micro-seismic events, and this class cluster mainly concentrated in the fault area of Yongshaba mine. The activity characteristics of micro-seismic events of this class cluster were analyzed, including 24-hour micro-seismic event distribution, activity rate, apparent volume, Schmidt number and stiffness coefficient. According to the variation characteristics of activity law, it was put forward that the point when the micro-seismic activity rate drops sharply and the cumulative apparent volume curve rises suddenly, and the Schmidt number and the stiffness coefficient rise first and then decrease afterwards can be taken as the early-warning point of destructive event of rock instability. The research on activity characteristics of micro-seismic events can provide effective predictive information for the analysis of destructive events, which plays an important role to ensure the work safety of mine.