Predicting Tunnel Squeezing Using the SVM-BP Combination Model

Rock squeezing has a large influence on tunnel construction safety; thus, when designing and constructing tunnels it is highly important to use reliable method for predicting from incomplete data. In this study, combination SVM-BP (support vector machine-back-propagation) model proposed classify the deformation caused by surrounding rock squeezing. We design different characteristic parameters three types of classifiers (a SVM model, BP model) tunnel-squeezing prediction experiments analyse accuracy predictions models influences results. contrast other methods, verified be reliable. The results show that four characteristics: diameter (D), buried depth (H), quality index (Q) support stiffness (K) reflect effect sufficiently classification. combines advantages both an neural network. It possesses flexible nonlinear modelling ability perform parallel processing large-scale information. Therefore, achieves better classification performance than do or separately. Moreover, coupling D, H, K significant impact predicted