地质参数分析与隧道超挖预测优化研究

发布时间：2018-06-18 01:55

本文选题：隧道 + 超挖　；参考：《现代隧道技术》2015年03期

【摘要】：隧道超挖会增加施工成本,有可能还会引发坍塌、大变形等问题。文章以明山隧道超挖控制工程为例,分析了其超挖特征,并构建了以爆破参数为试验常量、有效地质参数为输入量、实际超挖量为输出量的预测模型。通过采用线性(FDA)、非线性(CG)和智能预测(SVM)方法对隧道超挖进行了预测和判别比较,结果表明:3种预测方法的相关性系数(R2)分别为0.694,0.718和0.947;其中,FDA预测模型相关性系数最小,CG预测模型较FDA预测模型的相关性系数精度有微小提高,SVM预测模型具有最高的相关性系数,预测精度显著提高,且在数据变化突兀点仍能表现出较强的适应性;SVM预测模型可以实现高精度定量的超挖优化预测,而CG预测模型可以提供快速且精度可控的简单预测。
[Abstract]:Tunnel overexcavation will increase construction cost, may also lead to collapse, large deformation and other problems. Taking Mingshan tunnel overbreak control project as an example, the characteristics of overexcavation are analyzed, and a prediction model with blasting parameters as test constant, effective geological parameters as input and actual overexcavation as output is constructed. By using linear FDAG, nonlinear CGG) and intelligent prediction (SVM) method, the tunnel overbreak is predicted and compared. The results showed that the correlation coefficient (R2) of the three prediction methods was 0.694 and 0.947, respectively. The minimum correlation coefficient of the FDA prediction model and the CG prediction model had the highest correlation coefficient compared with that of the FDA prediction model, and the accuracy of the CG model was slightly higher than that of the FDA prediction model, and the correlation coefficient of the CG model was higher than that of the FDA prediction model. The accuracy of prediction is improved significantly, and the SVM prediction model can achieve high precision and quantitative overbreak optimization prediction, while CG prediction model can provide simple prediction with fast and controllable precision.
【作者单位】：中国地质大学(武汉)岩土钻掘与防护教育部工程研究中心;中国地质大学(武汉)工程学院;湖南大学土木工程学院;
【基金】：国家自然科学基金(41402259) 湖北省自然科学基金重点项目(2013CFA110) 中国地质大学(武汉)教学实验室开放基金(SKJ2013103)
【分类号】：U452.11;U456.3

【参考文献】