Cluster Analysis Of Monitoring Data Of Coal Mine Microseism

The development of mining activities to the depth destroys the original rock stress in the depth,which leads to the geological dynamic disaster.Rock burst is one of the typical geological dynamic disasters in coal mine production.With the increasing depth of coal mining,the geological dynamic disaster of coal mine is more and more frequent,and the impact of mine pressure is more and more serious.At present,the study of rock burst is still to be developed.In order to better predict and forewarn rock burst,it is particularly important to study the aggregation characteristics and activity rules of impact microseismic events.The research goal of this paper is to carry out intelligent spatiotemporal clustering analysis of microseismic monitoring data,to complete automatic and accurate division of time and space area of microseismic events,to realize the subarea preprocessing of microseismic monitoring data,to make the microseismic monitoring data better realize the follow-up early warning and prediction research on the basis of subarea preprocessing,so as to improve the accuracy of early warning and prediction of rock burst.Microseismic events are spatiotemporal events.Only through comprehensive investigation can we find closely related sets in time,space and other non spatial attributes,so as to achieve scientific and reasonable clustering.Based on the consideration of the characteristics of microseismic data and the comprehensive comparison of algorithms,this paper selects the spatiotemporal density clustering algorithm which is suitable for microseismic data,and combines the spatiotemporal characteristics of microseismic events to carry out the clustering research.Based on the limitation analysis and pertinence improvement of st-dbscan clustering algorithm,the initial parameter adaptive and heuristic optimization strategies are studied,and the parameter adaptive kst-dbscan algorithm based on microseisms is designed.The rationality of clustering results and the suitability of clustering division are determined by relative evaluation,internal index evaluation and physical analysis.Finally,the clustering validity is verified,and the maximum clustering cluster with the best parameters and the microseismic event of the mining face in the same period are selected to study the early warning index.The comparative study shows that the activity characteristics and distribution rules of microseismic events are more obvious,and the accuracy of the early warning and prediction of rockburst is higher through the pre-processing operation of clustering and zoning.Paper has 19 pictures,12 tables,and 84 references.