Intelligent prediction of ground deformation by small samples in the shield tunnel construction

Through learning from the observation of ground deformation in shield tunnel construction, the prediction model was established by choosing a suitable structure of ANN and relevant parameters. Small samples was trained for predicting behavior during subsequent construction. By using multi-step scrolling technology, the curve of ground deformation in recent days was obtained. The curves were drawn to reflect hogging or sagging condition in the tunnel direction. The method was used to the regional tunnel between Pudong South Station and Nanpu Bridge Station of the second project of rail transit "Bright Pearl Line" in Shanghai. It shows that the predicted results are in well accordance with the field observations according to the suggested method.