地下洞室超挖预测的距离判别分析方法和应用

Distance Discriminant Analysis Method and Its Application for Prediction of Over-excavation in Underground Opening

超挖问题是隧道开挖施工中不可避免的问题。本文基于距离判别分析理论并结合工程实际,建立地下洞室块体超挖预测的距离判别分析模型;根据工程实例中影响隧洞块体超挖的地质特征与组成,选用两组节理、一组层面的产状,以及洞室结构面的延伸和间距等指标作为洞室超挖预测的判别因子;以21组实测的洞室地质参数数据作为学习样本进行训练,建立相应线性判别函数预测超挖块体大小,经过训练后的模型回判估计误判率为0。最后,从野外地质调查着手,结合某一抽水蓄能电站的实测地质资料,应用建立的判别函数对平洞4(PD4)的超挖块体大小进行预测。结果表明,距离判别分析模型预测精度高,回代估计的误判率低,为地下洞室块体超挖预测提供了一种新思路。

Over-excavation is an inevitable fact in tunnel excavation. Based on the principles of the Mahalanobis Distance Discriminating Analysis （DDA）, a distance discriminating analysis model is proposed in this paper to predict over-excavation of underground opening. According to engineering practice, the main factors affecting overbreaking in tunnels include the geological characteristics of rock mass and the geological constitution, which are considered in the forecast DDA model. Five parameters influencing the over-excavation of opening, including two groups of joints, one group of layer surface, extension and space between structure faces, are selected as geometric parameters. Twenty-one sets of engineering data in an underground opening are used as the training samples. Corresponding discriminating functions are obtained to classify the samples to be discriminated. The re-substitution method is used to verify the Stability of the DDA model and the ratio of mistake-discrimination is equal to zero after the DDA model is trained. Some field survey data in a pump-storage hydroplant were used to test the discriminant ability of the DDA model. The results show that the prediction results are identical with the actual situation, with high predicted accuracy and low misdiscrimination rate. The DDA model can be used in practical engineering.