RISK PREDICTION MODEL OF ROCK-BURST BASED ON COMBINED ANN
Rock-burst is one of the coal and rock dynamical disasters that must be paid attention to when mining is carried out. At present, most risk prediction methods remain at the stage of simple statistical studies and the single factor forecast. In those methods, only the mining geology factors are considered, and the mining technical conditions are neglected. Therefore the predicted results cannot reach the desired precision. In this paper, the main influencing factors of the rock-burst risk both the mining geology factors and the mining technical conditions are comprehensively considered. Based on the sufficient existing rock-burst data, a combined artificial neural network (ANN) risk forecast model of rock-burst is proposed by utilizing the genetic algorithm to train back-propagation (BP) neural networks. It is capable for both the self-learning and strong robustness of neural networks and the global stochastically searching of genetic algorithms. Obviously, the effect of artificially determining the rock-burst risk indexes can be eliminated. The practical engineering application indicates that this method is not only precise, simple, but also intelligent. The predicted result by this method is based on the actual conditions. This model can be satisfactorily applied to the relevant engineering projects with quality results.
JIAN SUN LIAN-GUO WANG HUA-LEI ZHANG YI-FENG SHEN
State Key Laboratory for Geomechanics & Deep Underground Engineering, China University of Mining &Te State Key Laboratory for Geomechanics & Deep Underground Engineering, China University of Mining & T
国际会议
The 7th International Symposium on Rockburst and Seismicity in Mines(2009年第七届国际岩爆与微振动性学术研讨会)
大连
英文
1353-1360
2009-08-21（万方平台首次上网日期，不代表论文的发表时间）