Clustering And Early Warning Research On Coal Mine Microseismic Data Based On Earthquake Swarm Sequence

In the process of coal mining,various disasters are often accompanied,among which rock burst is one of the typical dynamic disasters.With the continuous extension of mine mining to the depths,the frequency of rock bursts has gradually increased,and the damage caused by them has become more and more serious.For the monitoring of rock bursts,the microseismic monitoring system is a widely used effective method.It can record the microseismic events occurring in the entire mine in real time continuously and dynamically,and save it as the spatio-temporal-energy microseismic monitoring data format.The microseismic monitoring data contains a large amount of disaster evolution information.Through in-depth analysis of it,the early warning of rock burst disasters can be effectively pre-warned and the safety production level of coal mines can be improved.Based on the analysis of the characteristics of microseismic monitoring data and the comparative study of clustering algorithms,the thesis designs and improves the existing spatio-temporal clustering algorithm to realize adaptive clustering of coal mine microseismic monitoring data,and then combines the coal mine field monitoring data and the early-warning indicators based on the sequence of earthquake swarms to achieve the feasibility analysis of clustering effect and the effectiveness verification of pre-warning.The details are as follows:In the part of the selection of clustering algorithm,on the one hand,in order to achieve accurate cluster division of earthquake swarms,the time and space of the data must be considered;on the other hand,the applicability of different spatiotemporal clustering algorithm must be considered.Therefore,combining the characteristics of microseismic data and the characteristics of different spatiotemporal clustering algorithms,the thesis chooses the spatiotemporal density clustering algorithm STDBSCAN as the basic algorithm for research.In the part of the design of improved clustering algorithm,the limitations of STDBSCAN are analyzed first,and the TST-DBSCAN clustering algorithm based on time window sorting is designed for the limitations of ST-DBSCAN,and then compare the clustering results of the TST-DBSCAN algorithm with the clustering results of the STDBSCAN algorithm through a random spatio-temporal data example.The comparison of clustering results proves the superiority of TST-DBSCAN.In the part of the feasibility and effectiveness of clustering,the TST-DBSCAN algorithm is used to cluster the microseismic data,and the optimal clustering result is determined through the parameter combination experiment and the clustering effect evaluation index.Select the largest earthquake swarm sequence in the result and the microseismic data of the same working face at the same time to calculate and analyze the early warning indicators.The comparative study shows that the early warning index calculated by the regional earthquake swarm sequence obtained by the TST-DBSCAN clustering algorithm is more accurate,which verifies the feasibility and effectiveness of the clustering.The thesis improves the ST-DBSCAN clustering algorithm,and designs the TSTDBSCAN algorithm based on time window sorting,which divides the microseismic monitoring data into time and space highly correlated earthquake swarm sequences to improve the accuracy of rock burst warning.It reduces disaster losses,improves the level of coal mine safety production,and has important theoretical and practical significance for the study of early warning,prediction and prevention of rock bursts.The thesis has 31 figures,9 tables and 94 references.