Prediction Model Of Grey BP Neural Network Based On Wavelet Transform In Deep Foundation Deformation Prediction Application

With China’s rapid development of urban construction, urban construction, the guide has gradually extended to the ground, especially in urban high-rise buildings, more and more subway construction are becoming more common, the scale and difficulty of deep excavation growing. Deep excavation in dense buildings between the excavation, will cause all kinds of deformation, this deformation is not only reflected in the Foundation’s own displacement, including settlement and tilt effect excavation around the building. Thus the establishment of appropriate model deformation monitoring is an important means to ensure the safety of excavation. Currently, deformation analysis and forecasting, there are various deformation analyzing and forecasting models, including single model such as a large number of theoretical gray system, BP neural networks, regression analysis, and the combination of a single model from the combination model.This selection of wavelet theory, gray model. BP neural network theory of the three models. Grey system theory to predict exponential change, and dealing with small data samples or incomplete information when the situation has outstanding effect, BP neural network, with good computing power and error correction capability features, combined with the advantages of both gray Construction neural network model by gray model to predict the prediction value obtained as the input sample value neural network, the actual observation data as the desired output, through training, the neural network and predict results.Based on the basis of BP Neural Network Combined Model on wavelet transform effectively weed out the original data series high-frequency noise, to retain the characteristics of useful information that gray BP neural network model based on wavelet optimization using MATLAB tool to achieve, through engineering examples settlement monitoring data, the horizontal displacement of the two data test data analysis and BP neural network, gray GM(1,1) model, wavelet optimization gray model of the three models predict the result of the comparison obtained wavelet optimization gray BP neural network processing excavation forecast noisy in reliability, can be a good application.