基于电磁辐射的隧道变形多参数模型及其应用

基于丽香铁路中义隧道的219组现场实测数据,开展了电磁辐射参数与隧道变形量、变形速率的相关性研究,选用电磁辐射强度、脉冲、能量指标中与隧道变形量、变形速率最相关的参数,建立了隧道变形量和变形速率的多参数模型,并利用该模型对开挖后隧道断面的变形演化规律进行了预测。结果表明：(1)电磁辐射强度极大峰值事件数占比、脉冲数、超过平均能量的峰值事件数占比分别是电磁辐射强度、脉冲、能量指标中与隧道变形量最相关的参数；(2)电磁辐射强度极大峰值事件数、脉冲数、超过平均能量的峰值事件数的波动速率分别是电磁辐射强度、脉冲、能量指标中与隧道变形速率最相关的参数；(3)采用偏最小二乘法(PLSR)建立了隧道变形量和变形速率的3参数PLSR模型,其计算值与实测值的相关系数分别为0.838和0.927,验证了该模型的有效性；(4)现场预测结果显示,3参数PLSR模型的预测值比一元回归模型更接近实测值,且预测可靠性更稳定,表明该模型能提高电磁辐射技术在工程实际中的现场应用水平。

Construction and Application of Multi-parameter Model for Tunnel Deformation Based on Electromagnetic Radiation

Based on 219 sets of field test data of Zhongyi Tunnel on Lixiang Railway, the correlation analysis between electromagnetic radiation (EMR) parameters and tunnel deformation (including deformation amount and deformation rate) were analyzed. The most relevant parameters among the intensity, pulse and energy indexes were used to develop the new multi-parameter models for deformation amount and deformation rate. The tunnel deformation amount and deformation rate after excavation were predicted by the developed models of this study. The results showed that: (1) Among the EMR intensity, pulse and energy indexes, the most relevant parameters of tunnel deformation amount are the percentage of maximum peak events of EMR intensity in total events, percentage of peak events which exceeding average energy in total events and pulse number, respectively. (2) The fluctuation rate of maximum peak events of EMR intensity, the fluctuation rate of peak events exceeding the average energy and the fluctuation rate of pulse number are the most relevant parameters of the tunnel deformation rate in the EMR intensity, pulse and energy indexes, respectively. (3) The three-parameter PLSR models of deformation amount and deformation rate were established by the Partial Least Squares Regression (PLSR) method. The correlation coefficients between the calculated values and the measured values of tunnel deformation amount and deformation rate were 0.838 and 0.927, respectively. It showed the effectiveness of these proposed models. (4) According to the flied application results, the prediction results by the three-parameter PLSR models of this study were closer to the field data than those by the unitary regression models. The prediction accuracy of the proposed models was more stable. It indicated that these proposed models can promote the field application of EMR technology in engineering practice.