深基坑变形影响因子分析及预测方法研究

Study on the Influence Factors Analysis and Prediction Method of Deep Foundation Pit Deformation

中文关键词:基坑变形预测  影响因子  BP神经网络  黄金分割法  层次分析法

英文关键词:deformation prediction of foundation pit  influence factors  BP neural network  fibonacci method  AHP

中文摘要:

      影响基坑变形的因子有很多，有主要因子和次要因子。根据实际工程经验以及相关资料，在较为全面地总结所有影响基坑变形因子的基础上，结合层次分析法，筛选出主要影响因子，建立一定的量化标准。将影响因子作为输入层，构建基于主要影响因子的BP神经网络。结合天津市中国铁建国际城1D地块深基坑工程，选出围护结构发生明显变形的各时间段，建立17×106的训练数据，采用黄金分割法对隐含层的节点数进行筛选，以4 m，8 m，12 m，16 m处的变形数据为目标层进行训练和仿真。最后，对其它测斜点的变形进行预测，精度满足要求，验证了这种影响因子选择和样本选择方法的有效性，对基坑变形预测有一定的应用指导意义。

英文摘要:

      There are many factors that affect the deformation of foundation pit, including major factors and secondary factors. According to the engineering experience and relevant information, based on a more comprehensive summary of the factors and identifying the major factors ,the quantitative criteria are established by combining the AHP. Using the factors as the input layers, the BP neural network can be constructed based on these factors. Taking the deep foundation pit engineering of Chinese Railway International City 1D block as an example, selecting the time periods that the retaining structure have deformations and establishing the 17×106 training data, the node numbers of hidden layer can be settled down by using the Fibonacci method. Then take the deformation data of 4 m,8 m,12 m,16 m as target layer to train and simulate. At last, using the network that have trained to make prediction of other points, the accuracy meets the requirements. This conclusion verifies the validity of this impact factor selection and sample selection method and has some guiding significance for the deformation predicting of deep foundation.