摘要
隧道拱顶沉降是多种因素共同作用的一个十分复杂的过程,很难用数学模型进行精确计算。实际施工过程中,围岩情况经常发生变化,为满足设计要求,必须严格控制隧道拱顶沉降。传统的预测模型都只是利用拱顶沉降监测数据建立单变量模型进行拟合并预测。隧道开挖过程中,拱顶变形所受影响因素较多,导致监测数据序列中常常出现离散型较大的数据,单一变量模型预测精度受这些离散数据的影响较大,而且筛除离散型较大的数据会直接影响模型预测精度,因此单一模型只能对拱顶沉降量做粗略的估计。针对这一问题,根据隧道变形的同时性和内在相关性,利用拱顶变形监测数据和同期两侧收敛变形监测数据构建带输入变量的时序模型、GM(1,2)模型和BP模型分别对拱顶沉降变形进行预测,并运用实例验证了所建模型的有效性。通过对两种模型的预测精度进行对比可知,单一变量的时序模型只能对变形的趋势作出预测,预测精度较低,难以对拱顶沉降进行有效预测,而单一变量的GM(1,1)预测模型则完全失效。为了避免单一模型自身的缺陷导致预测精度降低,同时使不同模型间优势互补,本研究建立了基于以上3种带输入变量模型的集成预测模型,其加权系数采用熵值法确定;最后将该模型运用于宝汉高速白庙子隧道中进行检验,结果表明该集成模型更有效,预测精度更高。
The settlement of tunnel vault is a very complicated process with combined actions of muhiple factors, which is difficult to calculate accurately with mathematical model. In the actual construction process, the surrounding rock often changes. In order to meet the design requirements, we must strictly control the tunnel vault settlement. The traditional prediction model only uses the vault settlement monitoring data to establish the univariate model to fit and predict. In the tunnel excavation process, there are many factors affect the deformation of vault. Therefore, the discrete data are often found in the monitoring data sequence, the prediction accuracy of the univariate model is greatly influenced by these discrete data, and screening out the more discrete data will directly affect the prediction accuracy of the model, so the single model can only make a rough estimation of the settlement of vault. To solve the problem, according to the simultaneity and intrinsic correlation of tunnel deformation, the time series model, GM （1,2） model and BP model with input variables is constructed by using vault deformation monitoring data and simultaneous convergence deformation monitoring data of 2 sides to predict the settlement of the vault respectively. In addition, the effectiveness of the model is verified by examples. By comparing the prediction accuracies of the 2 models, it is found that the univariate time series model can only predict the trend of deformation, the prediction accuracy is too low, which is difficult to effectively predict the vault settlement, and the univariate GM （ 1, 1 ） prediction model is completely ineffective. In order to avoid the lower prediction accuracy of single models due to their own defects, and make use of the complementary advantages among different models, an integrated prediction model based on the above 3 models is established, and its weighting coefficient is determined by entropy method. The proposed model has eventually been applied to Baimiaozi Tunnel which located
出处
《公路交通科技》
CAS
CSCD
北大核心
2017年第12期90-96,共7页
Journal of Highway and Transportation Research and Development
关键词
隧道工程
多变量模型
集成预测
拱顶沉降
灰色模型
tunnel engineering
multivariable model
integrated prediction
vault subsidence
gray model
