Research On Surrounding Rock Stability Forecasting Of Mined-out Area Based On SVM

In this thesis, surrounding rock stability of mined-out area is forecasted by support vector machine. The theory is a new application in data mining algorithm at this field. Combined similarity material simulation experiment and verification in site research with SVM, crack and destabilization of surrounding rock medium is predicted. It offered quantitative criteria for danger source discrimination and alarm during mining induced dynamic hazard, and provided credible guidance for in site timely and accurately predicted surrounding crack and destabilization of mined-out area.Firstly, the paper overviewed data mining algorithm at present. Secondly, SVM is introduced which is a new theory in data mining algorithm. Then, based on Support Vector Regression, we discussed regularization parameter C andε-insensitive zone selection. It is well known that SVM generalization performance (estimation accuracy) depends on a good setting of hyper-parameters C ,εand the kernel parameters. RBF (radial basis function) width parameter reflects the distribution/range of x-values of training data. So in this paper we discussed C ,εparameters selection for given RBF. Finally, under these assumptions, we obtained generalization parameters value selection by central limit theorems and triple standard deviation elements of probability statistics, then verified through cases and got good performance.Based on above works, SVM applied for similarity material simulation experiment of daliuta mine. We compared RBF with Sigmoid kernel functions during forecasting of surrounding rock stability. Results showed that the former is better than the latter, at the same time, result is given to illustrate selection reasonability of parametersε,C .At last, SVR is generalized during similarity material simulation experiment of huating mine, result is also well. SVM is valuable for generalization application.