地质参数分析与隧道超挖预测优化研究

Optimization of Tunnel Overbreak Prediction Based on Geological Parameter Analyses

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。

Considering that overbreak of rock may cause construction cost increases, large deformation or even tunnel collapse, and using the Mingshan tunnel as an example, this paper analyzes overbreak characteristics and establishes a prediction model. Using the blasting parameters as a constant, the effective geological parameters as an input and the actual overbreak volume as an output, three prediction methods are compared： the Fisher Discrimination Analysis Method（FDA）, the Conjugate Gradient Method（CG） and the Support Vector Machine Method（SVM）. The results show that their correlation coefficients R2 are 0.694, 0.718 and 0.947,with the correlation coefficient of the SVM model being the highest and the CG model coming second. The SVM model has sound prediction precision and adaptability even at the data point where abrupt change occurs. Adopting the SVM model can result in optimal quantitative prediction with high precision, while adopting the CG model can provide rapid and simple prediction with controllable precision.