【作者】 刘海燕；

【导师】 杨松林； 王斌；

【作者基本信息】 北京交通大学 ， 摄影测量与遥感， 2012， 硕士

【摘要】 随着城市建设的快速发展,城市空间的使用日益紧张,而地下空间的发展很好的缓解了此矛盾,深基坑工程也是其中之一,它作为承载建筑上部结构的部分,是一项很复杂且重要的工程,它的稳定与否不仅关系到上部结构的安全,还会对周围建筑物的变形产生影响,所以深基坑工程的变形研究受到越来越多学者的重视。基坑的变形受到很多因素的影响,它的变形很难用力学模型或者统一的经验公式来进行计算,本文将在前人研究的基础上,利用北京市某深基坑开挖过程中监测点的观测数据,以基坑开挖前期的变形数据来分析和预测后期的变形,并用实测数据验证预测的准确性,更好地反馈给设计施工,指导基坑开挖,避免施工过程中产生安全事故,同时也能有效的保护周围建筑物的稳定。本文引入灰色系统法和人工神经网络方法并详细分析了两种方法用于数据预测的基本理论,两种方法都能在一定程度上,在一定条件下对基坑的变形进行预测,但都存在缺点,灰色系统法通过对原始数据的整理来寻找规律,只适于呈指数增长的短期时间序列的变形预测。而利用人工神经网络方法进行预测,在监测数据序列很短的情况下,得出的结果误差相对较大。本文根据这种情况,将探讨利用灰色系统法能较精确预测短期变形的优点,与BP神经网络模型结合的新方法,基于北京市某深基坑工程的监测数据,利用Matlab软件,进行深基坑变形预测方法的分析,分别用灰色系统GM(1,1)模型和人工神经网络模型进行预测分析对比,然后将两种方法结合起来,用优化模型进行分析预测,结果说明优化模型对深基坑的变形预测结果具有更高的精度和准确性,同时,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 rapid development of city construction, the use of ur ban space increasingly nervous, while the development of underground space eas es the contradiction well. One kind of type is deep foundation pit engineering w hich is very complicated and important as the part of bearing the weight of the u pper structure, and its stable or not related to the safety of the upper structure, even had an impact on the deformation of the surrounding buildings, so its defor mation prediction receiving more and more attention of scholars.The pit deformation is influenced by so many factors that it is difficult to use the mechanical model or unified experience formula to compute the deformation. In order to guide the foundation pit excavation, avoiding safety accident in the process of constructing, protecting the stability of the surrounding buildings effectively, this paper will be based on the former research, use the observation data in the process of deep foundation pit of Beijing excavation of monitoring stations, with the early data to predict the late deformation of the foundation pit.This paper introduces the gray system method and artificial neural network method and detailed analysis the basic theory of these two methods are used in prediction. In some extent the two methods can do deformation forecast in certain conditions, while there are shortcomings. The gray system method looking for the sorting of law through the original data, it is suitable for prediction of the deformation of growing exponentially short-term time series. In the case of monitoring data sequence is very short, the prediction results of artificial neural network method are relatively large error.According to this kind of situation, this paper will be discussed a new method which combination with the advantage of the grey system method can be accurately predict short-term deformation and BP neural network model. Based on one foundation pit engineering monitoring data, do some analysis of foundation pit deformation forecast method. This paper based on the Matlab software, respectively for the gray system GM (1,1) model and artificial neural network model to forecast analysis contrast, and then combine with the two methods, using this optimization model to do analysis. The deep foundation pit deformation forecast results of the optimization model show that they are have higher precision sex and accuracy, using this model can forecast the trend of foundation pit deformation well, achieve good prediction effect, explain the suggested method has high reliability and applicability.更多还原