Study On Artificial Neural Network Method For Displacement Prediction In Foundation Pits

As the depth of foundation pits becomes bigger and bigger, the study of prediction and control of the displacement of foundation pits becomes one of the hot subjects studied in geotechnology field. The prediction of displacement of foundation pits is a complex and nonlinear problem, and the common-used methods, such as theoretical computation method and analysis method based on the monitored data, have their faults in solving this problem. So, exploring a feasible and practical theory of method to solve the problem of predicating the displacement of foundation pits has important theoretical and practical value.In this paper, the frame and the character of arguments, together with the training algorithm of artificial neural network are deeply analyzed. BP(Error Back Propagation) model is built up to predicate displacement of foundation pits. Then, this model is used to predicate the displacement of two foundation pits engineering projects. The predication results are identical with the monitored data. Comparing and analyzing the result predicated by BP model with the results predicated by GM model and AR model, it is illustrated that BP model has a better predication effect than those two models. So, it is testified that using BP model to predicate the displacements of foundation pits is feasible and effective. Also, BP model could be used in practical engineering projects.