关键词: 深基坑, 地下连续墙, 人工免疫识别, RBF神经网络, 变形预测

Abstract: Deep foundation pit is affected by several factors, and becomes the key problem in geotechnical engineering. The deformation must be controlled rigorously according to the actual situation surrounding environment and building safety grade. After analyzing the measured data of deformation, the prediction of deformation of supporting structure is made, and the safety of deep foundation pit is guranteed. To insure the construction quality of the deep foundation pit, deformation observation of the supporting structure is carried out in many deep pit in order to predict the development of deformation accurately during construction. A radial basis function (RBF) neural network learning algorithm based on immune recognition principle is studied. In the algorithm, the input data are regarded as antigens, and antibodies are regarded as the compression mapping of antigenas, i.e. the hidden layer centers. The weights of the output layer are determined by adopting the least square method, which can improve convergence speed and precision of the RBF neural network. A new time series prediction method based on artificial immune RBF neural network is also presented, and its application is discussed. The simulation experiment indicates that this method is effective in time series prediction.

Key words: deep foundation pit, diaphragm wall, artificial immune recognition, RBF neural network, deformation prediction