Abstract:

Prediction of the deformation is one of the most important methods for the construction parameter adjustment for deep foundation pit. However, it is still a chilling task to effectively predict accurate deformation in engineering application. We proposed deformation prediction model, which is based on the neural network optimized by particle swarm optimization, for the deformation of the deep foundation pit based on filed data. The proposed model is established by using the existing monitoring data as input parameters of neural network. The initial weights and threshold values of neural network model are optimized by using particle swarm optimization to improve the prediction accuracy and prediction efficiency of the neural network algorithm. The proposed method is used for the foundation pit located in north plaza of Changchun railway station comprehensive traffic transfer center. The results show that for the No.8 point measuring horizontal displacement, the root mean square error (RMSE) of the horizontal displacement of No.8 points is 3.78%, the mean absolute percentage error (MAPE) is 5.48%; for the No.9 point measuring ground settlement, the number respectively are 5.62% and 3.23%. Results show that the proposed method can be reliably used to predict the deformation of the deep foundation pit.

Key words: foundation pit, deformation prediction, particle swarm optimization, neural network