Research on Rockburst Classification Prediction Based on BP-SVM Model

Rockburst is a complex destabilization phenomenon which combination of multiple factors, the study rockburst for classification prediction can help prevent and control engineering geological hazards, reduce casualties property damage. To achieve efficient accurate solve problem propensity assessment, six evaluation factors are selected as rock explosion system: tangential stress σθ, uniaxial compressive strength σc, tensile strengthσt, to ratio σθ/σc(BCF), σc/σt (SCF), elastic deformation energy index Wet in this study. Widely collected domestic international groups data, 420 sets valid samples were obtained by data processing. Establish grading models based on BP neural networks support vector machines respectively, then establish BP-SVM arithmetic mean weights standard deviation weights, analyzing comparing rating results 120 among them. Accuracy, Precision, Recall, Specificity, F1 Score metrics evaluate performance different models, show that several obtain effective results, weight model proposed paper has best accuracy effect, better than traditional single machine learning method.