【作者】 王磊；

【导师】 唐平英；

【作者基本信息】 长沙理工大学 ， 大地测量学与测量工程， 2014， 硕士

【摘要】 随着社会不断的发展进步,城市土地开发的成本也在不断上升,开发商从以前单纯地追求增加楼体的高度,到现在开始到地下探寻更多可利用的空间,深基坑已不再是一个陌生的名词。基础是承载建筑上部结构的部分,它的稳定与否,既关系到上部结构的安全,还关系到周围建筑物的稳定,越来越多的学者和科研工作者加入到了基坑变形研究的队伍中,但影响基坑变形的因素有许多,已有的数学和力学模型不能直接用来研究变形的发展趋势,这往往需要研究者在数据海洋中找出他们的联系,然后建立数据模型。在最近的20年里,基坑施工技术在德国和其他发达国家已经发展成熟,产生了一种与传统的基坑开挖方式完全相反的新技术和新概念—逆作法施工技术,它可以缩短施工周期,降低基坑围护的成本。本论文通过研究武汉某逆作法深基坑工程,在探讨深基坑变形预测理论的基础上,对施工过程中监测数据进行采集、分析,用前几期观测数据建立模型,预测后几期变形,再利用后几期的实测数据进行模型修正,这样经过反复的预测、修正,最终得到一个较为完善的预测模型,经与后面各期实测变形比较,模型预测变形与实测变形具有较好的一致性,对我国类似基坑工程项目施工变形监测、预测具有参考作用。本人在建模中采用的数据分析方法是目前流行的灰色系统和人工神经网络方法。在一定程度上,两种方法都能够预测基坑的变形,但都有各自的缺点。灰色系统的缺点是,只适用于呈指数增长的变形数据,而人工神经网络法则在监测数据较少的情况下预测得出的结果误差相对比较大。基于以上两种方法的优缺点,本文建立了集灰色系统法和人工神经网络法的优点的优化组合模型。基于Matlab平台,使用模型进行深基坑变形预测,结果表明：优化组合模型对深基坑的变形预测结果具有更高的精度和准确性,特别对复杂条件下非线性的深基坑变形预测具有较高的可靠性和适用性。更多还原

【Abstract】 With the development of the society, the cost of developing the urban land is raising, in the past, land developers liked increasing the height of the building, but now they prefer to develop the depth of the land, deep foundation pit is no longer a strange word.The base is the bearing part of the upper structure construction, it related to the safety of the upper structure, and to the stability of the surrounding buildings as well, more and more scholars and scientists join the foundation pit deformation researches, but many factors affect the deformation of foundation pit, we cannot directly used the existing mathematical and mechanical model to study the development trend of deformation, which often require researchers to find this out in the complex data in a wide range of marine connection, then set up the data model.Reverse construction method is the emerging technologies of deep foundation pit construction in nearly 20 years, which have been widely used in Germany and other developed countries, reverse construction method is an opposite concept to the traditional technique of foundation pit construction, it can achieve the purpose of building the upper structure and the infrastructure construction at the same time, and it is the right choice of multistory basement construction in the future, it can not only shorten the period of construction, but also can reduce the cost of the enclosing foundation pit. In this paper, the actual background of the deep foundation pit engineering---Wuhan for one’s name on the basis of the deep foundation pit discovered in reverse construction method. I study on the exploration on the basis of the prediction searches of deep foundation pit deformation and joined the challengeable and uncertainable elements--- reverse construction method, using the actual deformation monitoring data to establish model, and in the continuous experiments, I modified the model and finally verified that the reverse construction method is full of superiority. This paper would provide guiding information to the other central China region of the loess soil foundation.In terms of data analysis, the data analysis theory which I used is gray system method and artificial neural network. Through the practice I founded that the two methods can predict the deformation of foundation pit in a certain extent, but they all have many disadvantages.The shortcomings of Gray system method is only applicable to exponential growth, but is not predictive for the index of the data, and the artificial neural network in the case of monitoring data sequence is very short and it is concluded that the results of the prediction error is relatively large.Based on the advantages and disadvantages of the above two methods, this paper will discuss the grey system method and the advantages of artificial neural network model. Using Matlab software, carries, on the deep foundation pit deformation prediction by gray system GM (1,1) model, artificial neural network model and comparing the comprehensive prediction model to forecast and the results, the results show that the optimization model has higher precision and accuracy.更多还原