基于岭回归-马氏距离的隧道围岩分类预测方法的研究

摘要

Based on ridge regression and Mahalanobis distance distinguish method ,the rock quality in-dex ,integrality index ,uniaxial compression strength ,longitudinal wave velocity ,elastic resistance coef-ficient and friction coefficient are determined by Mahalanobis distance distinguish method .Then the weight coefficients are determined by the ridge regression method ,and these factors that less impor-tant are taken out .Finally a model for classification of surrounding rock is established by the factors of the uniaxial compression strength ,elastic resistance coefficient and friction coefficient ,and it is verified by a certain tunnel in Guangdong Province .The results show that the ridge regression and Mahalanobis distance distinguish method are of excellent performance ,and their results are basically the same as the true .At last ,the model is applied to the Dongshen tunnel ,and the results are in good agreement with the hydraulic tunnel design specifications (SL279-2002) ,w hich proves that the mod-el used to classify the surrounding rock of tunnel is reasonable and effective .The successful applica-tion of this method can give references for classification of surrounding rock in other similar tunnel projects .%基于岭回归与马氏距离判别法的基本原理，选取岩石质量指标RQD、完整性系数Kv、单轴饱和抗压强度Rw、纵波波速Vp、弹性抗力系数Ko和结构面摩擦系数f作为评价指标，通过岭回归分析确定其权重系数，根据权重系数的大小剔除权重较小的指标，最终采用Rw，Ko和f建立马氏距离判别模型，并利用广东省某隧道工程进行模型检验。结果表明：岭回归-马氏距离判别法预测结果与实际相吻合，从而验证了岭回归-马氏距离判别法用于围岩分类预测是可行的。最后，运用建立的岭回归-马氏距离判别模型对东深供水改造工程隧洞围岩进行预测，经计算分析发现，所选洞段围岩判别结果与《水工隧洞设计规范》（SL279-2002）的分类结果一致，进一步证明了该方法用于隧道围岩分类预测的合理性及有效性。