支持向量机在围岩变形监测中的应用

根据支特向量机优越的非线性拟合性能，建立变形量的时间序列预测模型，滚动预测围岩变形量，提高了预测模型的训练速度和预测推广能力。该方法用于西乡-固戍盾构段围岩变形预测，并与BP神经网络预测进行比较。结果表明这种模型可预测区间较长且具有较高的准确度，能够科学地指导现场施工和监测。

Application of Support Vector Machine in Rock Deformation Monitoring

According to the powerful capacity of the support vector machine in non-linear modeling, the time serial forecast model is estab-lished to predict rock deformation continuously. The training and the predictive ability of the forecast model is promoted. The method is used in Xixiang-Gushu Shield segment to predict the rock deformation. Compared with BP neural network, the results show that a longer range and higher accuracy forecast interval can be predicted. So the site construction and monitoring can be guided scientifically by this model.