PREDICATION OF ROCKBURST BASED ON GAUSSIAN PROCESSES WITH THE COMBINATORIAL KERNEL FUNCTION
Rockburst is a kind of dynamic phenomenon for the surrounding rock mass in deep underground works. At present, it is still difficult to understand the mechanism of rockburst and carry out effective predication of rockburst. The Gaussian process is a completely new machine developed rapidly in recent years. Compared with ANN and SVM, this method has some advantages for example: easy programming, self-adaptive acquisition of hyperparameters, flexible non-parameters inference, prediction with probability interpretation and so on. Based on the analysis of main factors of rockburst, the Gaussian process with the combinatorial kernel function obtained by combination of squared exponential and rational quadratic covariance function is implemented by learning machine routine in Matlab for overcoming poor predictive precision and network generalization ability of single kernel function. Then the automatic relevant determination is introduced into combinatorial Gaussian kernel function in the programme and OP regression model with regard to hyperparameters was established. Meanwhile, the correlation and characteristics selection about inputs and prediction for testing samples on the basis of the net are completed respectively. The predicted results show that it is feasible and valid to use the GP regression model for predicating rockburst. Compared with RBF neural network and SVM, the prediction precision of GPR with the kernel function is relatively better.
JIA-QI GUO CHUN-SHENG QIAO CHONG XU LI-CHAO CHENG
School of Civil Engineering, Beijing Jiaotong University Beijing, 100044, P.R.China School of Energy Resource, Hebei university of Engineering Handan, 056038, P.R.China
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
1147-1152
2009-08-21（万方平台首次上网日期，不代表论文的发表时间）