基于SVM-BP模型非完整数据的隧道围岩挤压变形预测

Prediction of Tunnel Surrounding Rock Extrusion Deformation Based on SVM-BP Model with Incomplete Data

采用一种可靠的方法预测隧道围岩挤压变形对隧道的设计与施工至关重要。文章构建了用于围岩挤压变形分类的SVM-BP组合模型,通过设计不同特征参数和3种分类器“SVM模型、BP模型、SVM-BP组合模型”用于隧道挤压预测的试验,分析了不同模型的预测准确性和特征参数对预测结果的影响,验证了SVM-BP模型的可靠性。研究结果表明:采用隧道直径D、隧道埋深H、岩石质量指数Q和支撑刚度K这4个特征可较好地反映围岩挤压变形的分类效果;SVM-BP模型组合了SVM和BP神经网络模型的优点,具有灵活的非线性建模能力和大规模信息的并行处理能力,因此,SVM-BP模型比SVM和BP模型的分类性能更优;D,H和K这3个指标共同耦合对隧道围岩挤压变形预测结果的影响较大。

It is very important for the design and construction of tunnel to use a reliable method to predict the extrusion deformation of tunnel surrounding rock.In this paper,a SVM-BP combination model for classification of surrounding rock extrusion deformation is constructed.Different characteristic parameters and three classifiers“SVM model,BP model and SVM-BP combination model”are designed for tunnel extrusion prediction.The prediction accuracy of different models and the influence of characteristic parameters on prediction results are analyzed,and the reliability of SVM-BP model is verified.the results show that:The four characteristics of tunnel diameter D,tunnel buried depth H,rock quality index Q and support stiffness K can well reflect the classification effect of surrounding rock extrusion deformation;SVM-BP model combines the advantages of SVM and BP neural network model;it has flexible nonlinear modeling ability and parallel processing ability of large-scale information,therefore,SVM-BP model has better classification performance than that of SVM and BP model;the coupling of H,D and K has great influence on the prediction results of tunnel surrounding rock extrusion deformation.