Study of Characteristics of Mining Induced Seismicity in Deep Metal Mine and Rockburst Prediction Based on Support Vector Machine
|School||Central South University|
|Course||Underground Space Science and Engineering|
|Keywords||deep mining mining induced seismicity rockburst support vector machine prediction|
Mining induced seismicity and rockburst in deep metal mine becomes the important disasters which constrained mining and threatened the safety of underground structures, equipments and persons. So the prediction and control of mining induced seismicity and rockburst become an urgent problem to be solved.Combined with 973 subject "Blasting and excavation induced rockburst and fracture mechanism in deep hard rock", prediction of mining induced seismicity and rockburst are studied in deep mine by theoretical analysis and site monitoring, based on the data of micro-seismic monitoring in Dongguashan Copper Mine. The main contents and achievements of this paper are as follows:(1) Based on the analysis of several practical examples, some experience of analizing the stability of mine according to the change law of seismic parameters are summarized. Such as when accumulated volume increased suddenly while schmidt decreased at the same time, rockburst almost is inevitable.(2) The stress distribution on the spatial shows that there is stress concentration in rib pillar between two panels. Thus it is the dangerous place which should be paid more attention to and some appropriate preventive measures should be taken for safe production.(3) The distribution characteristics of mining induced seismicity in time are analized, which furtherly proving that mining induced seismicity have close relation to actual production. The activity of mining induced seismicity whose magnitude is less than 0 during a day has 3 peak periods, respectively 6:00-8:00,14:00-16:00, and 22:00-24:00. This is consistent with the time of blasting,8:00,16:00,24:00, though it is a little ahead of that which shows the foreshocks phenomenon.(4) The influence of excavation and filling to the stability of the mine is studied through analyzing characteristic change under different conditions. When the stope is empty, the incidence of mining induced seismicity is very high, while after filling the incidence reduced significantly. So the stability of stope has a close connection to the state of goaf and filling. The stope should be filled as timely as possible after excavation, which is beneficial to the mine stability.(5) The prediction of rockburst proneness is researched by support vector machine. On a summary of various influence factors of rockburst, four attributes which are seismic moment, stress drop and elastic energy index’Wet’andσc/σt are selected as the influence factors, with energy release rate as the forecast goal to characterize the size of rockburst proneness. The prediction model which is suitable for Dongguashan Copper Mine is established by using the new proposed support vector machine method. The results show that the prediction accuracy is high.(6) With the same data, the prediction accuracy based on BP network is lower than that of support vector machine, especially when the data is small, which proved that support vector machine is a better intelligence method, having a more broad prospect of application.